Update files from the datasets library (from 1.4.0)

Release notes: https://github.com/huggingface/datasets/releases/tag/1.4.0

.gitattributes

@@ -0,0 +1,27 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bin.* filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zstandard filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,225 @@
+---
+annotations_creators:
+- expert-generated
+language_creators:
+- crowdsourced
+- expert-generated
+languages:
+- "fr"
+- "de"
+- "es"
+- "ca"
+- "it"
+- "ru"
+- "zh-CN"
+- "pt"
+- "fa"
+- "et"
+- "mn"
+- "nl"
+- "tr"
+- "ar"
+- "sv-SE"
+- "lv"
+- "sl"
+- "ta"
+- "ja"
+- "id"
+- "cy"
+licenses:
+- cc-by-nc-4-0
+multilinguality:
+- multilingual
+size_categories:
+- 100K<n<1M
+source_datasets:
+- extended|other-common-voice
+task_categories:
+- other
+task_ids:
+- other-other-speech-translation
+---
+
+# Dataset Card for covost2
+
+## Table of Contents
+- [Dataset Description](#dataset-description)
+  - [Dataset Summary](#dataset-summary)
+  - [Supported Tasks](#supported-tasks-and-leaderboards)
+  - [Languages](#languages)
+- [Dataset Structure](#dataset-structure)
+  - [Data Instances](#data-instances)
+  - [Data Fields](#data-instances)
+  - [Data Splits](#data-instances)
+- [Dataset Creation](#dataset-creation)
+  - [Curation Rationale](#curation-rationale)
+  - [Source Data](#source-data)
+  - [Annotations](#annotations)
+  - [Personal and Sensitive Information](#personal-and-sensitive-information)
+- [Considerations for Using the Data](#considerations-for-using-the-data)
+  - [Social Impact of Dataset](#social-impact-of-dataset)
+  - [Discussion of Biases](#discussion-of-biases)
+  - [Other Known Limitations](#other-known-limitations)
+- [Additional Information](#additional-information)
+  - [Dataset Curators](#dataset-curators)
+  - [Licensing Information](#licensing-information)
+  - [Citation Information](#citation-information)
+  - [Contributions](#contributions)
+
+## Dataset Description
+
+- **Homepage:** https://github.com/facebookresearch/covost
+- **Repository:** https://github.com/facebookresearch/covost
+- **Paper:** https://arxiv.org/abs/2007.10310
+- **Leaderboard:** [Needs More Information]
+- **Point of Contact:** Changhan Wang (changhan@fb.com), Juan Miguel Pino (juancarabina@fb.com), Jiatao Gu (jgu@fb.com)
+
+### Dataset Summary
+
+CoVoST 2 is a large-scale multilingual speech translation corpus covering translations from 21 languages into English \
+and from English into 15 languages. The dataset is created using Mozillas open-source Common Voice database of \
+crowdsourced voice recordings. There are 2,900 hours of speech represented in the corpus.
+
+### Supported Tasks and Leaderboards
+
+`speech-translation`: The dataset can be used for Speech-to-text translation (ST). The model is presented with an audio file in one language and asked to transcribe the audio file to written text in another language. The most common evaluation metric is the BLEU score. Examples can be found at https://github.com/pytorch/fairseq/blob/master/examples/speech_to_text/docs/covost_example.md .
+
+### Languages
+
+The dataset contains the audio, transcriptions, and translations in the following languages, French, German, Dutch, Russian, Spanish, Italian, Turkish, Persian, Swedish, Mongolian, Chinese, Welsh, Catalan, Slovenian, Estonian, Indonesian, Arabic, Tamil, Portuguese, Latvian, and Japanese. 
+
+## Dataset Structure
+
+### Data Instances
+
+A typical data point comprises the path to the audio file, usually called `file`, its transcription, called `sentence`, and the translation in target language called `translation`.
+
+```
+{'client_id': 'd277a1f3904ae00b09b73122b87674e7c2c78e08120721f37b5577013ead08d1ea0c053ca5b5c2fb948df2c81f27179aef2c741057a17249205d251a8fe0e658',
+ 'file': '/home/suraj/projects/fairseq_s2t/covst/dataset/en/clips/common_voice_en_18540003.mp3',
+ 'id': 'common_voice_en_18540003',
+ 'sentence': 'When water is scarce, avoid wasting it.',
+ 'translation': 'Wenn Wasser knapp ist, verschwenden Sie es nicht.'}
+```
+
+### Data Fields
+
+- file: A path to the downloaded audio file in .mp3 format.
+
+- sentence: The transcription of the audio file in source language.
+
+- translation: The transcription of the audio file in the target language. 
+
+- id: unique id of the data sample.
+
+### Data Splits
+
+| config   | train  | validation | test  |
+|----------|--------|------------|-------|
+| en_de    | 289430 | 15531      | 15531 |
+| en_tr    | 289430 | 15531      | 15531 |
+| en_fa    | 289430 | 15531      | 15531 |
+| en_sv-SE | 289430 | 15531      | 15531 |
+| en_mn    | 289430 | 15531      | 15531 |
+| en_zh-CN | 289430 | 15531      | 15531 |
+| en_cy    | 289430 | 15531      | 15531 |
+| en_ca    | 289430 | 15531      | 15531 |
+| en_sl    | 289430 | 15531      | 15531 |
+| en_et    | 289430 | 15531      | 15531 |
+| en_id    | 289430 | 15531      | 15531 |
+| en_ar    | 289430 | 15531      | 15531 |
+| en_ta    | 289430 | 15531      | 15531 |
+| en_lv    | 289430 | 15531      | 15531 |
+| en_ja    | 289430 | 15531      | 15531 |
+| fr_en    | 207374 | 14760      | 14760 |
+| de_en    | 127834 | 13511      | 13511 |
+| es_en    | 79015  | 13221      | 13221 |
+| ca_en    | 95854  | 12730      | 12730 |
+| it_en    | 31698  | 8940       | 8951  |
+| ru_en    | 12112  | 6110       | 6300  |
+| zh-CN_en | 7085   | 4843       | 4898  |
+| pt_en    | 9158   | 3318       | 4023  |
+| fa_en    | 53949  | 3445       | 3445  |
+| et_en    | 1782   | 1576       | 1571  |
+| mn_en    | 2067   | 1761       | 1759  |
+| nl_en    | 7108   | 1699       | 1699  |
+| tr_en    | 3966   | 1624       | 1629  |
+| ar_en    | 2283   | 1758       | 1695  |
+| sv-SE_en | 2160   | 1349       | 1595  |
+| lv_en    | 2337   | 1125       | 1629  |
+| sl_en    | 1843   | 509        | 360   |
+| ta_en    | 1358   | 384        | 786   |
+| ja_en    | 1119   | 635        | 684   |
+| id_en    | 1243   | 792        | 844   |
+| cy_en    | 1241   | 690        | 690   |
+
+
+## Dataset Creation
+
+### Curation Rationale
+
+[Needs More Information]
+
+### Source Data
+
+#### Initial Data Collection and Normalization
+
+[Needs More Information]
+
+#### Who are the source language producers?
+
+[Needs More Information]
+
+### Annotations
+
+#### Annotation process
+
+[Needs More Information]
+
+#### Who are the annotators?
+
+[Needs More Information]
+
+### Personal and Sensitive Information
+
+[Needs More Information]
+
+## Considerations for Using the Data
+
+### Social Impact of Dataset
+
+[Needs More Information]
+
+### Discussion of Biases
+
+[Needs More Information]
+
+### Other Known Limitations
+
+[Needs More Information]
+
+## Additional Information
+
+### Dataset Curators
+
+[Needs More Information]
+
+### Licensing Information
+
+cc-by-nc-4.0
+
+### Citation Information
+
+```
+@misc{wang2020covost,
+    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},
+    author={Changhan Wang and Anne Wu and Juan Pino},
+    year={2020},
+    eprint={2007.10310},
+    archivePrefix={arXiv},
+    primaryClass={cs.CL}
+```
+
+### Contributions
+
+Thanks to [@patil-suraj](https://github.com/patil-suraj) for adding this dataset.

covost2.py

@@ -0,0 +1,195 @@
+# coding=utf-8
+# Copyright 2021 The TensorFlow Datasets Authors and the HuggingFace Datasets Authors.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+# Lint as: python3
+
+import csv
+import os
+
+import pandas as pd
+
+import datasets
+
+
+_VERSION = "1.0.0"
+
+_CITATION = """
+@misc{wang2020covost,
+    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},
+    author={Changhan Wang and Anne Wu and Juan Pino},
+    year={2020},
+    eprint={2007.10310},
+    archivePrefix={arXiv},
+    primaryClass={cs.CL}
+"""
+
+_DESCRIPTION = """
+CoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English \
+and from English into 15 languages. The dataset is created using Mozilla’s open source Common Voice database of \
+crowdsourced voice recordings.
+
+Note that in order to limit the required storage for preparing this dataset, the audio
+is stored in the .mp3 format and is not converted to a float32 array. To convert, the audio
+file to a float32 array, please make use of the `.map()` function as follows:
+
+
+```python
+import torchaudio
+
+def map_to_array(batch):
+    speech_array, _ = torchaudio.load(batch["file"])
+    batch["speech"] = speech_array.numpy()
+    return batch
+
+dataset = dataset.map(map_to_array, remove_columns=["file"])
+```
+"""
+
+_HOMEPAGE = "https://github.com/facebookresearch/covost"
+
+# fmt: off
+XX_EN_LANGUAGES = ["fr", "de", "es", "ca", "it", "ru", "zh-CN", "pt", "fa", "et", "mn", "nl", "tr", "ar", "sv-SE", "lv", "sl", "ta", "ja", "id", "cy"]
+EN_XX_LANGUAGES = ["de", "tr", "fa", "sv-SE", "mn", "zh-CN", "cy", "ca", "sl", "et", "id", "ar", "ta", "lv", "ja"]
+# fmt: on
+
+COVOST_URL_TEMPLATE = "https://dl.fbaipublicfiles.com/covost/covost_v2.{src_lang}_{tgt_lang}.tsv.tar.gz"
+
+
+def _get_builder_configs():
+    builder_configs = [
+        datasets.BuilderConfig(name=f"en_{lang}", version=datasets.Version(_VERSION)) for lang in EN_XX_LANGUAGES
+    ]
+
+    builder_configs += [
+        datasets.BuilderConfig(name=f"{lang}_en", version=datasets.Version(_VERSION)) for lang in XX_EN_LANGUAGES
+    ]
+    return builder_configs
+
+
+class Covost2(datasets.GeneratorBasedBuilder):
+    """CoVOST2 Dataset."""
+
+    VERSION = datasets.Version(_VERSION)
+
+    BUILDER_CONFIGS = _get_builder_configs()
+
+    @property
+    def manual_download_instructions(self):
+        return """\
+        You should download the Common Voice v4 dataset from https://commonvoice.mozilla.org/en/datasets.
+        and unpack it to a path `{COVOST_ROOT}/{SOURCE_LANG_ID}` and then pass the `{COVOST_ROOT}` path as `data_dir`
+        via `datasets.load_dataset('covost2', data_dir="path/to/covost_root")`
+        """
+
+    def _info(self):
+        return datasets.DatasetInfo(
+            description=_DESCRIPTION,
+            features=datasets.Features(
+                client_id=datasets.Value("string"),
+                file=datasets.Value("string"),
+                sentence=datasets.Value("string"),
+                translation=datasets.Value("string"),
+                id=datasets.Value("string"),
+            ),
+            supervised_keys=("file", "translation"),
+            homepage=_HOMEPAGE,
+            citation=_CITATION,
+        )
+
+    def _split_generators(self, dl_manager):
+        data_root = os.path.abspath(os.path.expanduser(dl_manager.manual_dir))
+        source_lang, target_lang = self.config.name.split("_")
+        source_path = os.path.join(data_root, source_lang)
+
+        if not os.path.exists(source_path):
+            raise FileNotFoundError(
+                "{} does not exist. Make sure you insert a manual dir via "
+                "`datasets.load_dataset('covost2', data_dir=...)` that includes files uncompressed files from the "
+                "COVOST2 archive. Manual download instructions: {}".format(
+                    data_root, self.manual_download_instructions
+                )
+            )
+
+        covost_url = COVOST_URL_TEMPLATE.format(src_lang=source_lang, tgt_lang=target_lang)
+        extracted_path = dl_manager.download_and_extract(covost_url)
+
+        covost_tsv_path = os.path.join(extracted_path, f"covost_v2.{source_lang}_{target_lang}.tsv")
+        cv_tsv_path = os.path.join(source_path, "validated.tsv")
+
+        return [
+            datasets.SplitGenerator(
+                name=datasets.Split.TRAIN,
+                gen_kwargs={
+                    "source_path": source_path,
+                    "covost_tsv_path": covost_tsv_path,
+                    "cv_tsv_path": cv_tsv_path,
+                    "split": "train",
+                },
+            ),
+            datasets.SplitGenerator(
+                name=datasets.Split.VALIDATION,
+                gen_kwargs={
+                    "source_path": source_path,
+                    "covost_tsv_path": covost_tsv_path,
+                    "cv_tsv_path": cv_tsv_path,
+                    "split": "dev",
+                },
+            ),
+            datasets.SplitGenerator(
+                name=datasets.Split.TEST,
+                gen_kwargs={
+                    "source_path": source_path,
+                    "covost_tsv_path": covost_tsv_path,
+                    "cv_tsv_path": cv_tsv_path,
+                    "split": "test",
+                },
+            ),
+        ]
+
+    def _generate_examples(self, source_path, covost_tsv_path, cv_tsv_path, split):
+        covost_tsv = self._load_df_from_tsv(covost_tsv_path)
+        cv_tsv = self._load_df_from_tsv(cv_tsv_path)
+
+        df = pd.merge(
+            left=cv_tsv[["path", "sentence", "client_id"]],
+            right=covost_tsv[["path", "translation", "split"]],
+            how="inner",
+            on="path",
+        )
+
+        if split == "train":
+            df = df[(df["split"] == "train") | (df["split"] == f"train_covost")]
+        else:
+            df = df[df["split"] == split]
+
+        for i, row in df.iterrows():
+            yield i, {
+                "id": row["path"].replace(".mp3", ""),
+                "client_id": row["client_id"],
+                "sentence": row["sentence"],
+                "translation": row["translation"],
+                "file": os.path.join(source_path, "clips", row["path"]),
+            }
+
+    def _load_df_from_tsv(self, path):
+        return pd.read_csv(
+            path,
+            sep="\t",
+            header=0,
+            encoding="utf-8",
+            escapechar="\\",
+            quoting=csv.QUOTE_NONE,
+            na_filter=False,
+        )

dataset_infos.json

@@ -0,0 +1 @@
+{"en_de": {"description": "\nCoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla\u2019s open source Common Voice database of crowdsourced voice recordings.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .mp3 format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport torchaudio\n\ndef map_to_array(batch):\n    speech_array, _ = torchaudio.load(batch[\"file\"])\n    batch[\"speech\"] = speech_array.numpy()\n    return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "\n@misc{wang2020covost,\n    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},\n    author={Changhan Wang and Anne Wu and Juan Pino},\n    year={2020},\n    eprint={2007.10310},\n    archivePrefix={arXiv},\n    primaryClass={cs.CL}\n", "homepage": "https://github.com/facebookresearch/covost", "license": "", "features": {"client_id": {"dtype": "string", "id": null, "_type": "Value"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "translation": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "translation"}, "builder_name": "covost2", "config_name": "en_de", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 110716293, "num_examples": 289430, "dataset_name": "covost2"}, "validation": {"name": "validation", "num_bytes": 5971731, "num_examples": 15531, "dataset_name": "covost2"}, "test": {"name": "test", "num_bytes": 5689684, "num_examples": 15531, "dataset_name": "covost2"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/covost/covost_v2.en_de.tsv.tar.gz": {"num_bytes": 25779505, "checksum": "f414705b6197cc9a30cf05f8fc4471c8e99809a2c1f2c1da24a7ba617f42fce2"}}, "download_size": 25779505, "post_processing_size": null, "dataset_size": 122377708, "size_in_bytes": 148157213}, "en_tr": {"description": "\nCoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla\u2019s open source Common Voice database of crowdsourced voice recordings.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .mp3 format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport torchaudio\n\ndef map_to_array(batch):\n    speech_array, _ = torchaudio.load(batch[\"file\"])\n    batch[\"speech\"] = speech_array.numpy()\n    return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "\n@misc{wang2020covost,\n    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},\n    author={Changhan Wang and Anne Wu and Juan Pino},\n    year={2020},\n    eprint={2007.10310},\n    archivePrefix={arXiv},\n    primaryClass={cs.CL}\n", "homepage": "https://github.com/facebookresearch/covost", "license": "", "features": {"client_id": {"dtype": "string", "id": null, "_type": "Value"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "translation": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "translation"}, "builder_name": "covost2", "config_name": "en_tr", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 109474265, "num_examples": 289430, "dataset_name": "covost2"}, "validation": {"name": "validation", "num_bytes": 5914622, "num_examples": 15531, "dataset_name": "covost2"}, "test": {"name": "test", "num_bytes": 5619271, "num_examples": 15531, "dataset_name": "covost2"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/covost/covost_v2.en_tr.tsv.tar.gz": {"num_bytes": 23659131, "checksum": "9cfb1533cd2226c6ed13fe9462b2ecc5ebc9bb50fc0eb7d0dfe8a68ee4873d42"}}, "download_size": 23659131, "post_processing_size": null, "dataset_size": 121008158, "size_in_bytes": 144667289}, "en_fa": {"description": "\nCoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla\u2019s open source Common Voice database of crowdsourced voice recordings.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .mp3 format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport torchaudio\n\ndef map_to_array(batch):\n    speech_array, _ = torchaudio.load(batch[\"file\"])\n    batch[\"speech\"] = speech_array.numpy()\n    return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "\n@misc{wang2020covost,\n    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},\n    author={Changhan Wang and Anne Wu and Juan Pino},\n    year={2020},\n    eprint={2007.10310},\n    archivePrefix={arXiv},\n    primaryClass={cs.CL}\n", "homepage": "https://github.com/facebookresearch/covost", "license": "", "features": {"client_id": {"dtype": "string", "id": null, "_type": "Value"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "translation": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "translation"}, "builder_name": "covost2", "config_name": "en_fa", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 119490720, "num_examples": 289430, "dataset_name": "covost2"}, "validation": {"name": "validation", "num_bytes": 6423535, "num_examples": 15531, "dataset_name": "covost2"}, "test": {"name": "test", "num_bytes": 6103617, "num_examples": 15531, "dataset_name": "covost2"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/covost/covost_v2.en_fa.tsv.tar.gz": {"num_bytes": 26148420, "checksum": "d255d629f604aa32ea06dbb612acf7602221bb0e05ad0d652ad0cf9ba6894532"}}, "download_size": 26148420, "post_processing_size": null, "dataset_size": 132017872, "size_in_bytes": 158166292}, "en_sv-SE": {"description": "\nCoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla\u2019s open source Common Voice database of crowdsourced voice recordings.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .mp3 format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport torchaudio\n\ndef map_to_array(batch):\n    speech_array, _ = torchaudio.load(batch[\"file\"])\n    batch[\"speech\"] = speech_array.numpy()\n    return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "\n@misc{wang2020covost,\n    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},\n    author={Changhan Wang and Anne Wu and Juan Pino},\n    year={2020},\n    eprint={2007.10310},\n    archivePrefix={arXiv},\n    primaryClass={cs.CL}\n", "homepage": "https://github.com/facebookresearch/covost", "license": "", "features": {"client_id": {"dtype": "string", "id": null, "_type": "Value"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "translation": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "translation"}, "builder_name": "covost2", "config_name": "en_sv-SE", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 108557530, "num_examples": 289430, "dataset_name": "covost2"}, "validation": {"name": "validation", "num_bytes": 5845918, "num_examples": 15531, "dataset_name": "covost2"}, "test": {"name": "test", "num_bytes": 5580039, "num_examples": 15531, "dataset_name": "covost2"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/covost/covost_v2.en_sv-SE.tsv.tar.gz": {"num_bytes": 23671482, "checksum": "0c591c50b3b13db4df9365802c2cf8a69c3db00a39da6306fc44e76a1b25b4b6"}}, "download_size": 23671482, "post_processing_size": null, "dataset_size": 119983487, "size_in_bytes": 143654969}, "en_mn": {"description": "\nCoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla\u2019s open source Common Voice database of crowdsourced voice recordings.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .mp3 format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport torchaudio\n\ndef map_to_array(batch):\n    speech_array, _ = torchaudio.load(batch[\"file\"])\n    batch[\"speech\"] = speech_array.numpy()\n    return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "\n@misc{wang2020covost,\n    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},\n    author={Changhan Wang and Anne Wu and Juan Pino},\n    year={2020},\n    eprint={2007.10310},\n    archivePrefix={arXiv},\n    primaryClass={cs.CL}\n", "homepage": "https://github.com/facebookresearch/covost", "license": "", "features": {"client_id": {"dtype": "string", "id": null, "_type": "Value"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "translation": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "translation"}, "builder_name": "covost2", "config_name": "en_mn", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 123950136, "num_examples": 289430, "dataset_name": "covost2"}, "validation": {"name": "validation", "num_bytes": 6693044, "num_examples": 15531, "dataset_name": "covost2"}, "test": {"name": "test", "num_bytes": 6293633, "num_examples": 15531, "dataset_name": "covost2"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/covost/covost_v2.en_mn.tsv.tar.gz": {"num_bytes": 27527436, "checksum": "7ef1b63adb5cb6bbc7a1f3cfa7c2ceab9a83d99aa4727ba894d101b7b320beb7"}}, "download_size": 27527436, "post_processing_size": null, "dataset_size": 136936813, "size_in_bytes": 164464249}, "en_zh-CN": {"description": "\nCoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla\u2019s open source Common Voice database of crowdsourced voice recordings.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .mp3 format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport torchaudio\n\ndef map_to_array(batch):\n    speech_array, _ = torchaudio.load(batch[\"file\"])\n    batch[\"speech\"] = speech_array.numpy()\n    return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "\n@misc{wang2020covost,\n    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},\n    author={Changhan Wang and Anne Wu and Juan Pino},\n    year={2020},\n    eprint={2007.10310},\n    archivePrefix={arXiv},\n    primaryClass={cs.CL}\n", "homepage": "https://github.com/facebookresearch/covost", "license": "", "features": {"client_id": {"dtype": "string", "id": null, "_type": "Value"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "translation": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "translation"}, "builder_name": "covost2", "config_name": "en_zh-CN", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 106490939, "num_examples": 289430, "dataset_name": "covost2"}, "validation": {"name": "validation", "num_bytes": 5735331, "num_examples": 15531, "dataset_name": "covost2"}, "test": {"name": "test", "num_bytes": 5487808, "num_examples": 15531, "dataset_name": "covost2"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/covost/covost_v2.en_zh-CN.tsv.tar.gz": {"num_bytes": 24280932, "checksum": "7509b7c587326ef094c1e639111aef13f3888d26f707672dadec4d37d8e501d8"}}, "download_size": 24280932, "post_processing_size": null, "dataset_size": 117714078, "size_in_bytes": 141995010}, "en_cy": {"description": "\nCoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla\u2019s open source Common Voice database of crowdsourced voice recordings.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .mp3 format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport torchaudio\n\ndef map_to_array(batch):\n    speech_array, _ = torchaudio.load(batch[\"file\"])\n    batch[\"speech\"] = speech_array.numpy()\n    return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "\n@misc{wang2020covost,\n    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},\n    author={Changhan Wang and Anne Wu and Juan Pino},\n    year={2020},\n    eprint={2007.10310},\n    archivePrefix={arXiv},\n    primaryClass={cs.CL}\n", "homepage": "https://github.com/facebookresearch/covost", "license": "", "features": {"client_id": {"dtype": "string", "id": null, "_type": "Value"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "translation": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "translation"}, "builder_name": "covost2", "config_name": "en_cy", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 109317182, "num_examples": 289430, "dataset_name": "covost2"}, "validation": {"name": "validation", "num_bytes": 5894579, "num_examples": 15531, "dataset_name": "covost2"}, "test": {"name": "test", "num_bytes": 5626428, "num_examples": 15531, "dataset_name": "covost2"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/covost/covost_v2.en_cy.tsv.tar.gz": {"num_bytes": 24224499, "checksum": "72173329cec05fc70efbf9efa2144a3cc4fbbf3c2e59c01d176a8960f40d0da7"}}, "download_size": 24224499, "post_processing_size": null, "dataset_size": 120838189, "size_in_bytes": 145062688}, "en_ca": {"description": "\nCoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla\u2019s open source Common Voice database of crowdsourced voice recordings.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .mp3 format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport torchaudio\n\ndef map_to_array(batch):\n    speech_array, _ = torchaudio.load(batch[\"file\"])\n    batch[\"speech\"] = speech_array.numpy()\n    return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "\n@misc{wang2020covost,\n    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},\n    author={Changhan Wang and Anne Wu and Juan Pino},\n    year={2020},\n    eprint={2007.10310},\n    archivePrefix={arXiv},\n    primaryClass={cs.CL}\n", "homepage": "https://github.com/facebookresearch/covost", "license": "", "features": {"client_id": {"dtype": "string", "id": null, "_type": "Value"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "translation": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "translation"}, "builder_name": "covost2", "config_name": "en_ca", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 109922455, "num_examples": 289430, "dataset_name": "covost2"}, "validation": {"name": "validation", "num_bytes": 5924345, "num_examples": 15531, "dataset_name": "covost2"}, "test": {"name": "test", "num_bytes": 5623227, "num_examples": 15531, "dataset_name": "covost2"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/covost/covost_v2.en_ca.tsv.tar.gz": {"num_bytes": 24167201, "checksum": "efbcd8fcc730ce3900adb274d0842ad65980a559de0a840d4a8bcbe94eabf774"}}, "download_size": 24167201, "post_processing_size": null, "dataset_size": 121470027, "size_in_bytes": 145637228}, "en_sl": {"description": "\nCoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla\u2019s open source Common Voice database of crowdsourced voice recordings.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .mp3 format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport torchaudio\n\ndef map_to_array(batch):\n    speech_array, _ = torchaudio.load(batch[\"file\"])\n    batch[\"speech\"] = speech_array.numpy()\n    return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "\n@misc{wang2020covost,\n    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},\n    author={Changhan Wang and Anne Wu and Juan Pino},\n    year={2020},\n    eprint={2007.10310},\n    archivePrefix={arXiv},\n    primaryClass={cs.CL}\n", "homepage": "https://github.com/facebookresearch/covost", "license": "", "features": {"client_id": {"dtype": "string", "id": null, "_type": "Value"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "translation": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "translation"}, "builder_name": "covost2", "config_name": "en_sl", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 107987860, "num_examples": 289430, "dataset_name": "covost2"}, "validation": {"name": "validation", "num_bytes": 5838299, "num_examples": 15531, "dataset_name": "covost2"}, "test": {"name": "test", "num_bytes": 5537805, "num_examples": 15531, "dataset_name": "covost2"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/covost/covost_v2.en_sl.tsv.tar.gz": {"num_bytes": 23421999, "checksum": "4af4f68bbc89527f0cce69c99721a27ae13ec98faaba38436b742ad3a9999393"}}, "download_size": 23421999, "post_processing_size": null, "dataset_size": 119363964, "size_in_bytes": 142785963}, "en_et": {"description": "\nCoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla\u2019s open source Common Voice database of crowdsourced voice recordings.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .mp3 format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport torchaudio\n\ndef map_to_array(batch):\n    speech_array, _ = torchaudio.load(batch[\"file\"])\n    batch[\"speech\"] = speech_array.numpy()\n    return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "\n@misc{wang2020covost,\n    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},\n    author={Changhan Wang and Anne Wu and Juan Pino},\n    year={2020},\n    eprint={2007.10310},\n    archivePrefix={arXiv},\n    primaryClass={cs.CL}\n", "homepage": "https://github.com/facebookresearch/covost", "license": "", "features": {"client_id": {"dtype": "string", "id": null, "_type": "Value"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "translation": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "translation"}, "builder_name": "covost2", "config_name": "en_et", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 107707024, "num_examples": 289430, "dataset_name": "covost2"}, "validation": {"name": "validation", "num_bytes": 5810185, "num_examples": 15531, "dataset_name": "covost2"}, "test": {"name": "test", "num_bytes": 5543309, "num_examples": 15531, "dataset_name": "covost2"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/covost/covost_v2.en_et.tsv.tar.gz": {"num_bytes": 23223843, "checksum": "4966b7215adf6d29ad5d9c91030025b245c79c460594c0a0c8fbc2a95916e0ab"}}, "download_size": 23223843, "post_processing_size": null, "dataset_size": 119060518, "size_in_bytes": 142284361}, "en_id": {"description": "\nCoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla\u2019s open source Common Voice database of crowdsourced voice recordings.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .mp3 format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport torchaudio\n\ndef map_to_array(batch):\n    speech_array, _ = torchaudio.load(batch[\"file\"])\n    batch[\"speech\"] = speech_array.numpy()\n    return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "\n@misc{wang2020covost,\n    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},\n    author={Changhan Wang and Anne Wu and Juan Pino},\n    year={2020},\n    eprint={2007.10310},\n    archivePrefix={arXiv},\n    primaryClass={cs.CL}\n", "homepage": "https://github.com/facebookresearch/covost", "license": "", "features": {"client_id": {"dtype": "string", "id": null, "_type": "Value"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "translation": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "translation"}, "builder_name": "covost2", "config_name": "en_id", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 109456930, "num_examples": 289430, "dataset_name": "covost2"}, "validation": {"name": "validation", "num_bytes": 5896953, "num_examples": 15531, "dataset_name": "covost2"}, "test": {"name": "test", "num_bytes": 5634939, "num_examples": 15531, "dataset_name": "covost2"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/covost/covost_v2.en_id.tsv.tar.gz": {"num_bytes": 22904065, "checksum": "f2948f3473fafe6b437e8b55949b059d21da5b9893bd011f6d650a6bcba98f79"}}, "download_size": 22904065, "post_processing_size": null, "dataset_size": 120988822, "size_in_bytes": 143892887}, "en_ar": {"description": "\nCoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla\u2019s open source Common Voice database of crowdsourced voice recordings.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .mp3 format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport torchaudio\n\ndef map_to_array(batch):\n    speech_array, _ = torchaudio.load(batch[\"file\"])\n    batch[\"speech\"] = speech_array.numpy()\n    return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "\n@misc{wang2020covost,\n    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},\n    author={Changhan Wang and Anne Wu and Juan Pino},\n    year={2020},\n    eprint={2007.10310},\n    archivePrefix={arXiv},\n    primaryClass={cs.CL}\n", "homepage": "https://github.com/facebookresearch/covost", "license": "", "features": {"client_id": {"dtype": "string", "id": null, "_type": "Value"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "translation": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "translation"}, "builder_name": "covost2", "config_name": "en_ar", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 116732296, "num_examples": 289430, "dataset_name": "covost2"}, "validation": {"name": "validation", "num_bytes": 6280190, "num_examples": 15531, "dataset_name": "covost2"}, "test": {"name": "test", "num_bytes": 5947069, "num_examples": 15531, "dataset_name": "covost2"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/covost/covost_v2.en_ar.tsv.tar.gz": {"num_bytes": 25301304, "checksum": "cd780ad364b8b246b16cf76e4d8e14e21ca29e70b247a10856a923736e973122"}}, "download_size": 25301304, "post_processing_size": null, "dataset_size": 128959555, "size_in_bytes": 154260859}, "en_ta": {"description": "\nCoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla\u2019s open source Common Voice database of crowdsourced voice recordings.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .mp3 format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport torchaudio\n\ndef map_to_array(batch):\n    speech_array, _ = torchaudio.load(batch[\"file\"])\n    batch[\"speech\"] = speech_array.numpy()\n    return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "\n@misc{wang2020covost,\n    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},\n    author={Changhan Wang and Anne Wu and Juan Pino},\n    year={2020},\n    eprint={2007.10310},\n    archivePrefix={arXiv},\n    primaryClass={cs.CL}\n", "homepage": "https://github.com/facebookresearch/covost", "license": "", "features": {"client_id": {"dtype": "string", "id": null, "_type": "Value"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "translation": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "translation"}, "builder_name": "covost2", "config_name": "en_ta", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 146318684, "num_examples": 289430, "dataset_name": "covost2"}, "validation": {"name": "validation", "num_bytes": 7944020, "num_examples": 15531, "dataset_name": "covost2"}, "test": {"name": "test", "num_bytes": 7411400, "num_examples": 15531, "dataset_name": "covost2"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/covost/covost_v2.en_ta.tsv.tar.gz": {"num_bytes": 30037790, "checksum": "e95519f0157a779315bf8f75496f71a3fae7be5fec0375ca8fb217d6f1aac6fa"}}, "download_size": 30037790, "post_processing_size": null, "dataset_size": 161674104, "size_in_bytes": 191711894}, "en_lv": {"description": "\nCoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla\u2019s open source Common Voice database of crowdsourced voice recordings.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .mp3 format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport torchaudio\n\ndef map_to_array(batch):\n    speech_array, _ = torchaudio.load(batch[\"file\"])\n    batch[\"speech\"] = speech_array.numpy()\n    return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "\n@misc{wang2020covost,\n    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},\n    author={Changhan Wang and Anne Wu and Juan Pino},\n    year={2020},\n    eprint={2007.10310},\n    archivePrefix={arXiv},\n    primaryClass={cs.CL}\n", "homepage": "https://github.com/facebookresearch/covost", "license": "", "features": {"client_id": {"dtype": "string", "id": null, "_type": "Value"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "translation": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "translation"}, "builder_name": "covost2", "config_name": "en_lv", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 109532576, "num_examples": 289430, "dataset_name": "covost2"}, "validation": {"name": "validation", "num_bytes": 5905197, "num_examples": 15531, "dataset_name": "covost2"}, "test": {"name": "test", "num_bytes": 5625189, "num_examples": 15531, "dataset_name": "covost2"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/covost/covost_v2.en_lv.tsv.tar.gz": {"num_bytes": 24573927, "checksum": "b3bbbdf9ab66df79a00724145d43669305460c345bce950f0f2e252900b8bd53"}}, "download_size": 24573927, "post_processing_size": null, "dataset_size": 121062962, "size_in_bytes": 145636889}, "en_ja": {"description": "\nCoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla\u2019s open source Common Voice database of crowdsourced voice recordings.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .mp3 format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport torchaudio\n\ndef map_to_array(batch):\n    speech_array, _ = torchaudio.load(batch[\"file\"])\n    batch[\"speech\"] = speech_array.numpy()\n    return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "\n@misc{wang2020covost,\n    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},\n    author={Changhan Wang and Anne Wu and Juan Pino},\n    year={2020},\n    eprint={2007.10310},\n    archivePrefix={arXiv},\n    primaryClass={cs.CL}\n", "homepage": "https://github.com/facebookresearch/covost", "license": "", "features": {"client_id": {"dtype": "string", "id": null, "_type": "Value"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "translation": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "translation"}, "builder_name": "covost2", "config_name": "en_ja", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 114741253, "num_examples": 289430, "dataset_name": "covost2"}, "validation": {"name": "validation", "num_bytes": 6161930, "num_examples": 15531, "dataset_name": "covost2"}, "test": {"name": "test", "num_bytes": 5883608, "num_examples": 15531, "dataset_name": "covost2"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/covost/covost_v2.en_ja.tsv.tar.gz": {"num_bytes": 26664247, "checksum": "2308c0b1737cc3ed7662d7e1a748b29da71ea9023681a617bd51a81c11889671"}}, "download_size": 26664247, "post_processing_size": null, "dataset_size": 126786791, "size_in_bytes": 153451038}, "fr_en": {"description": "\nCoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla\u2019s open source Common Voice database of crowdsourced voice recordings.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .mp3 format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport torchaudio\n\ndef map_to_array(batch):\n    speech_array, _ = torchaudio.load(batch[\"file\"])\n    batch[\"speech\"] = speech_array.numpy()\n    return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "\n@misc{wang2020covost,\n    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},\n    author={Changhan Wang and Anne Wu and Juan Pino},\n    year={2020},\n    eprint={2007.10310},\n    archivePrefix={arXiv},\n    primaryClass={cs.CL}\n", "homepage": "https://github.com/facebookresearch/covost", "license": "", "features": {"client_id": {"dtype": "string", "id": null, "_type": "Value"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "translation": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "translation"}, "builder_name": "covost2", "config_name": "fr_en", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 75792665, "num_examples": 207374, "dataset_name": "covost2"}, "validation": {"name": "validation", "num_bytes": 5487082, "num_examples": 14760, "dataset_name": "covost2"}, "test": {"name": "test", "num_bytes": 5525498, "num_examples": 14760, "dataset_name": "covost2"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/covost/covost_v2.fr_en.tsv.tar.gz": {"num_bytes": 7282129, "checksum": "596d42accc976c607e7b4bd53903a0a03d5036af3b3ea276d7ff188981bb5b2f"}}, "download_size": 7282129, "post_processing_size": null, "dataset_size": 86805245, "size_in_bytes": 94087374}, "de_en": {"description": "\nCoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla\u2019s open source Common Voice database of crowdsourced voice recordings.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .mp3 format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport torchaudio\n\ndef map_to_array(batch):\n    speech_array, _ = torchaudio.load(batch[\"file\"])\n    batch[\"speech\"] = speech_array.numpy()\n    return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "\n@misc{wang2020covost,\n    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},\n    author={Changhan Wang and Anne Wu and Juan Pino},\n    year={2020},\n    eprint={2007.10310},\n    archivePrefix={arXiv},\n    primaryClass={cs.CL}\n", "homepage": "https://github.com/facebookresearch/covost", "license": "", "features": {"client_id": {"dtype": "string", "id": null, "_type": "Value"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "translation": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "translation"}, "builder_name": "covost2", "config_name": "de_en", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 47678171, "num_examples": 127834, "dataset_name": "covost2"}, "validation": {"name": "validation", "num_bytes": 5106253, "num_examples": 13511, "dataset_name": "covost2"}, "test": {"name": "test", "num_bytes": 5066500, "num_examples": 13511, "dataset_name": "covost2"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/covost/covost_v2.de_en.tsv.tar.gz": {"num_bytes": 9926797, "checksum": "a64efc469ba7bd392de32538acf49ff05ac7fe83dd705b9a2a07573e63a2fedd"}}, "download_size": 9926797, "post_processing_size": null, "dataset_size": 57850924, "size_in_bytes": 67777721}, "es_en": {"description": "\nCoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla\u2019s open source Common Voice database of crowdsourced voice recordings.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .mp3 format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport torchaudio\n\ndef map_to_array(batch):\n    speech_array, _ = torchaudio.load(batch[\"file\"])\n    batch[\"speech\"] = speech_array.numpy()\n    return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "\n@misc{wang2020covost,\n    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},\n    author={Changhan Wang and Anne Wu and Juan Pino},\n    year={2020},\n    eprint={2007.10310},\n    archivePrefix={arXiv},\n    primaryClass={cs.CL}\n", "homepage": "https://github.com/facebookresearch/covost", "license": "", "features": {"client_id": {"dtype": "string", "id": null, "_type": "Value"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "translation": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "translation"}, "builder_name": "covost2", "config_name": "es_en", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 29152515, "num_examples": 79015, "dataset_name": "covost2"}, "validation": {"name": "validation", "num_bytes": 4974593, "num_examples": 13221, "dataset_name": "covost2"}, "test": {"name": "test", "num_bytes": 4983920, "num_examples": 13221, "dataset_name": "covost2"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/covost/covost_v2.es_en.tsv.tar.gz": {"num_bytes": 3202080, "checksum": "aafb7255c00444cbff44ca92a947e16c8f7297f326b332f78b427f4f80240b97"}}, "download_size": 3202080, "post_processing_size": null, "dataset_size": 39111028, "size_in_bytes": 42313108}, "ca_en": {"description": "\nCoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla\u2019s open source Common Voice database of crowdsourced voice recordings.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .mp3 format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport torchaudio\n\ndef map_to_array(batch):\n    speech_array, _ = torchaudio.load(batch[\"file\"])\n    batch[\"speech\"] = speech_array.numpy()\n    return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "\n@misc{wang2020covost,\n    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},\n    author={Changhan Wang and Anne Wu and Juan Pino},\n    year={2020},\n    eprint={2007.10310},\n    archivePrefix={arXiv},\n    primaryClass={cs.CL}\n", "homepage": "https://github.com/facebookresearch/covost", "license": "", "features": {"client_id": {"dtype": "string", "id": null, "_type": "Value"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "translation": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "translation"}, "builder_name": "covost2", "config_name": "ca_en", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 35902579, "num_examples": 95854, "dataset_name": "covost2"}, "validation": {"name": "validation", "num_bytes": 4798435, "num_examples": 12730, "dataset_name": "covost2"}, "test": {"name": "test", "num_bytes": 4804941, "num_examples": 12730, "dataset_name": "covost2"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/covost/covost_v2.ca_en.tsv.tar.gz": {"num_bytes": 5021926, "checksum": "acbdbc2b365ac2d192f802b2df2b4796070044e6782193bd0eca5d08e514def1"}}, "download_size": 5021926, "post_processing_size": null, "dataset_size": 45505955, "size_in_bytes": 50527881}, "it_en": {"description": "\nCoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla\u2019s open source Common Voice database of crowdsourced voice recordings.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .mp3 format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport torchaudio\n\ndef map_to_array(batch):\n    speech_array, _ = torchaudio.load(batch[\"file\"])\n    batch[\"speech\"] = speech_array.numpy()\n    return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "\n@misc{wang2020covost,\n    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},\n    author={Changhan Wang and Anne Wu and Juan Pino},\n    year={2020},\n    eprint={2007.10310},\n    archivePrefix={arXiv},\n    primaryClass={cs.CL}\n", "homepage": "https://github.com/facebookresearch/covost", "license": "", "features": {"client_id": {"dtype": "string", "id": null, "_type": "Value"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "translation": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "translation"}, "builder_name": "covost2", "config_name": "it_en", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 11952709, "num_examples": 31698, "dataset_name": "covost2"}, "validation": {"name": "validation", "num_bytes": 3393315, "num_examples": 8940, "dataset_name": "covost2"}, "test": {"name": "test", "num_bytes": 3412207, "num_examples": 8951, "dataset_name": "covost2"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/covost/covost_v2.it_en.tsv.tar.gz": {"num_bytes": 1691247, "checksum": "da63a3ddf926a595971412e8fdf702405945ea6f9c6ac9de652a94fd6713183d"}}, "download_size": 1691247, "post_processing_size": null, "dataset_size": 18758231, "size_in_bytes": 20449478}, "ru_en": {"description": "\nCoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla\u2019s open source Common Voice database of crowdsourced voice recordings.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .mp3 format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport torchaudio\n\ndef map_to_array(batch):\n    speech_array, _ = torchaudio.load(batch[\"file\"])\n    batch[\"speech\"] = speech_array.numpy()\n    return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "\n@misc{wang2020covost,\n    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},\n    author={Changhan Wang and Anne Wu and Juan Pino},\n    year={2020},\n    eprint={2007.10310},\n    archivePrefix={arXiv},\n    primaryClass={cs.CL}\n", "homepage": "https://github.com/facebookresearch/covost", "license": "", "features": {"client_id": {"dtype": "string", "id": null, "_type": "Value"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "translation": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "translation"}, "builder_name": "covost2", "config_name": "ru_en", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 5610194, "num_examples": 12112, "dataset_name": "covost2"}, "validation": {"name": "validation", "num_bytes": 2819414, "num_examples": 6110, "dataset_name": "covost2"}, "test": {"name": "test", "num_bytes": 2923961, "num_examples": 6300, "dataset_name": "covost2"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/covost/covost_v2.ru_en.tsv.tar.gz": {"num_bytes": 1443078, "checksum": "4d23d3d771234ef1e4afd0fb62231d3e9490cebb4c445997715607feef806fec"}}, "download_size": 1443078, "post_processing_size": null, "dataset_size": 11353569, "size_in_bytes": 12796647}, "zh-CN_en": {"description": "\nCoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla\u2019s open source Common Voice database of crowdsourced voice recordings.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .mp3 format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport torchaudio\n\ndef map_to_array(batch):\n    speech_array, _ = torchaudio.load(batch[\"file\"])\n    batch[\"speech\"] = speech_array.numpy()\n    return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "\n@misc{wang2020covost,\n    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},\n    author={Changhan Wang and Anne Wu and Juan Pino},\n    year={2020},\n    eprint={2007.10310},\n    archivePrefix={arXiv},\n    primaryClass={cs.CL}\n", "homepage": "https://github.com/facebookresearch/covost", "license": "", "features": {"client_id": {"dtype": "string", "id": null, "_type": "Value"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "translation": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "translation"}, "builder_name": "covost2", "config_name": "zh-CN_en", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 2791288, "num_examples": 7085, "dataset_name": "covost2"}, "validation": {"name": "validation", "num_bytes": 1918796, "num_examples": 4843, "dataset_name": "covost2"}, "test": {"name": "test", "num_bytes": 1908633, "num_examples": 4898, "dataset_name": "covost2"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/covost/covost_v2.zh-CN_en.tsv.tar.gz": {"num_bytes": 587550, "checksum": "7a49bba59f8a05b68385e686a878332fb66080c12d4d85dc02b10f3f935447a4"}}, "download_size": 587550, "post_processing_size": null, "dataset_size": 6618717, "size_in_bytes": 7206267}, "pt_en": {"description": "\nCoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla\u2019s open source Common Voice database of crowdsourced voice recordings.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .mp3 format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport torchaudio\n\ndef map_to_array(batch):\n    speech_array, _ = torchaudio.load(batch[\"file\"])\n    batch[\"speech\"] = speech_array.numpy()\n    return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "\n@misc{wang2020covost,\n    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},\n    author={Changhan Wang and Anne Wu and Juan Pino},\n    year={2020},\n    eprint={2007.10310},\n    archivePrefix={arXiv},\n    primaryClass={cs.CL}\n", "homepage": "https://github.com/facebookresearch/covost", "license": "", "features": {"client_id": {"dtype": "string", "id": null, "_type": "Value"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "translation": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "translation"}, "builder_name": "covost2", "config_name": "pt_en", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 3095722, "num_examples": 9158, "dataset_name": "covost2"}, "validation": {"name": "validation", "num_bytes": 1133404, "num_examples": 3318, "dataset_name": "covost2"}, "test": {"name": "test", "num_bytes": 1384251, "num_examples": 4023, "dataset_name": "covost2"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/covost/covost_v2.pt_en.tsv.tar.gz": {"num_bytes": 476419, "checksum": "944cf077d47a0ee360fcde34ff42a77f4e898a6446c26ca9402ff543ef91e63b"}}, "download_size": 476419, "post_processing_size": null, "dataset_size": 5613377, "size_in_bytes": 6089796}, "fa_en": {"description": "\nCoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla\u2019s open source Common Voice database of crowdsourced voice recordings.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .mp3 format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport torchaudio\n\ndef map_to_array(batch):\n    speech_array, _ = torchaudio.load(batch[\"file\"])\n    batch[\"speech\"] = speech_array.numpy()\n    return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "\n@misc{wang2020covost,\n    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},\n    author={Changhan Wang and Anne Wu and Juan Pino},\n    year={2020},\n    eprint={2007.10310},\n    archivePrefix={arXiv},\n    primaryClass={cs.CL}\n", "homepage": "https://github.com/facebookresearch/covost", "license": "", "features": {"client_id": {"dtype": "string", "id": null, "_type": "Value"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "translation": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "translation"}, "builder_name": "covost2", "config_name": "fa_en", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 18015738, "num_examples": 53949, "dataset_name": "covost2"}, "validation": {"name": "validation", "num_bytes": 1241531, "num_examples": 3445, "dataset_name": "covost2"}, "test": {"name": "test", "num_bytes": 1263271, "num_examples": 3445, "dataset_name": "covost2"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/covost/covost_v2.fa_en.tsv.tar.gz": {"num_bytes": 3864623, "checksum": "1882aea149ce2ae9c553c41c2b7b6c837c4fc648a1a381705107e91679198e17"}}, "download_size": 3864623, "post_processing_size": null, "dataset_size": 20520540, "size_in_bytes": 24385163}, "et_en": {"description": "\nCoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla\u2019s open source Common Voice database of crowdsourced voice recordings.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .mp3 format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport torchaudio\n\ndef map_to_array(batch):\n    speech_array, _ = torchaudio.load(batch[\"file\"])\n    batch[\"speech\"] = speech_array.numpy()\n    return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "\n@misc{wang2020covost,\n    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},\n    author={Changhan Wang and Anne Wu and Juan Pino},\n    year={2020},\n    eprint={2007.10310},\n    archivePrefix={arXiv},\n    primaryClass={cs.CL}\n", "homepage": "https://github.com/facebookresearch/covost", "license": "", "features": {"client_id": {"dtype": "string", "id": null, "_type": "Value"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "translation": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "translation"}, "builder_name": "covost2", "config_name": "et_en", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 808508, "num_examples": 1782, "dataset_name": "covost2"}, "validation": {"name": "validation", "num_bytes": 690694, "num_examples": 1576, "dataset_name": "covost2"}, "test": {"name": "test", "num_bytes": 685375, "num_examples": 1571, "dataset_name": "covost2"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/covost/covost_v2.et_en.tsv.tar.gz": {"num_bytes": 246569, "checksum": "e0beadd6013b74efc2642c2289a67a0a3243bdbc5dc470798440b7538207eba0"}}, "download_size": 246569, "post_processing_size": null, "dataset_size": 2184577, "size_in_bytes": 2431146}, "mn_en": {"description": "\nCoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla\u2019s open source Common Voice database of crowdsourced voice recordings.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .mp3 format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport torchaudio\n\ndef map_to_array(batch):\n    speech_array, _ = torchaudio.load(batch[\"file\"])\n    batch[\"speech\"] = speech_array.numpy()\n    return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "\n@misc{wang2020covost,\n    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},\n    author={Changhan Wang and Anne Wu and Juan Pino},\n    year={2020},\n    eprint={2007.10310},\n    archivePrefix={arXiv},\n    primaryClass={cs.CL}\n", "homepage": "https://github.com/facebookresearch/covost", "license": "", "features": {"client_id": {"dtype": "string", "id": null, "_type": "Value"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "translation": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "translation"}, "builder_name": "covost2", "config_name": "mn_en", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 900588, "num_examples": 2067, "dataset_name": "covost2"}, "validation": {"name": "validation", "num_bytes": 765543, "num_examples": 1761, "dataset_name": "covost2"}, "test": {"name": "test", "num_bytes": 762577, "num_examples": 1759, "dataset_name": "covost2"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/covost/covost_v2.mn_en.tsv.tar.gz": {"num_bytes": 189710, "checksum": "9bf7633e24a9742211aeeebaee931d22e04a7e91a87e40d4f4957d408aa3dff4"}}, "download_size": 189710, "post_processing_size": null, "dataset_size": 2428708, "size_in_bytes": 2618418}, "nl_en": {"description": "\nCoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla\u2019s open source Common Voice database of crowdsourced voice recordings.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .mp3 format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport torchaudio\n\ndef map_to_array(batch):\n    speech_array, _ = torchaudio.load(batch[\"file\"])\n    batch[\"speech\"] = speech_array.numpy()\n    return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "\n@misc{wang2020covost,\n    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},\n    author={Changhan Wang and Anne Wu and Juan Pino},\n    year={2020},\n    eprint={2007.10310},\n    archivePrefix={arXiv},\n    primaryClass={cs.CL}\n", "homepage": "https://github.com/facebookresearch/covost", "license": "", "features": {"client_id": {"dtype": "string", "id": null, "_type": "Value"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "translation": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "translation"}, "builder_name": "covost2", "config_name": "nl_en", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 2468140, "num_examples": 7108, "dataset_name": "covost2"}, "validation": {"name": "validation", "num_bytes": 594458, "num_examples": 1699, "dataset_name": "covost2"}, "test": {"name": "test", "num_bytes": 594979, "num_examples": 1699, "dataset_name": "covost2"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/covost/covost_v2.nl_en.tsv.tar.gz": {"num_bytes": 543795, "checksum": "1cd5baf7282e6556d7463d26ae28cc717a64b9b7875bd1b38f226e04498b2336"}}, "download_size": 543795, "post_processing_size": null, "dataset_size": 3657577, "size_in_bytes": 4201372}, "tr_en": {"description": "\nCoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla\u2019s open source Common Voice database of crowdsourced voice recordings.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .mp3 format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport torchaudio\n\ndef map_to_array(batch):\n    speech_array, _ = torchaudio.load(batch[\"file\"])\n    batch[\"speech\"] = speech_array.numpy()\n    return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "\n@misc{wang2020covost,\n    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},\n    author={Changhan Wang and Anne Wu and Juan Pino},\n    year={2020},\n    eprint={2007.10310},\n    archivePrefix={arXiv},\n    primaryClass={cs.CL}\n", "homepage": "https://github.com/facebookresearch/covost", "license": "", "features": {"client_id": {"dtype": "string", "id": null, "_type": "Value"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "translation": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "translation"}, "builder_name": "covost2", "config_name": "tr_en", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 1391148, "num_examples": 3966, "dataset_name": "covost2"}, "validation": {"name": "validation", "num_bytes": 566458, "num_examples": 1624, "dataset_name": "covost2"}, "test": {"name": "test", "num_bytes": 570760, "num_examples": 1629, "dataset_name": "covost2"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/covost/covost_v2.tr_en.tsv.tar.gz": {"num_bytes": 280904, "checksum": "23130548e8ff9433c74b2193e0bc39d04d2aacb811841f592a7b35d358fad697"}}, "download_size": 280904, "post_processing_size": null, "dataset_size": 2528366, "size_in_bytes": 2809270}, "ar_en": {"description": "\nCoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla\u2019s open source Common Voice database of crowdsourced voice recordings.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .mp3 format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport torchaudio\n\ndef map_to_array(batch):\n    speech_array, _ = torchaudio.load(batch[\"file\"])\n    batch[\"speech\"] = speech_array.numpy()\n    return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "\n@misc{wang2020covost,\n    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},\n    author={Changhan Wang and Anne Wu and Juan Pino},\n    year={2020},\n    eprint={2007.10310},\n    archivePrefix={arXiv},\n    primaryClass={cs.CL}\n", "homepage": "https://github.com/facebookresearch/covost", "license": "", "features": {"client_id": {"dtype": "string", "id": null, "_type": "Value"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "translation": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "translation"}, "builder_name": "covost2", "config_name": "ar_en", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 743065, "num_examples": 2283, "dataset_name": "covost2"}, "validation": {"name": "validation", "num_bytes": 575077, "num_examples": 1758, "dataset_name": "covost2"}, "test": {"name": "test", "num_bytes": 552356, "num_examples": 1695, "dataset_name": "covost2"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/covost/covost_v2.ar_en.tsv.tar.gz": {"num_bytes": 109802, "checksum": "a8f7f8e393cc78aea1382e021bd6adf4ff75ee4ae96cddd96c80291fcc8c87cf"}}, "download_size": 109802, "post_processing_size": null, "dataset_size": 1870498, "size_in_bytes": 1980300}, "sv-SE_en": {"description": "\nCoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla\u2019s open source Common Voice database of crowdsourced voice recordings.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .mp3 format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport torchaudio\n\ndef map_to_array(batch):\n    speech_array, _ = torchaudio.load(batch[\"file\"])\n    batch[\"speech\"] = speech_array.numpy()\n    return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "\n@misc{wang2020covost,\n    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},\n    author={Changhan Wang and Anne Wu and Juan Pino},\n    year={2020},\n    eprint={2007.10310},\n    archivePrefix={arXiv},\n    primaryClass={cs.CL}\n", "homepage": "https://github.com/facebookresearch/covost", "license": "", "features": {"client_id": {"dtype": "string", "id": null, "_type": "Value"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "translation": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "translation"}, "builder_name": "covost2", "config_name": "sv-SE_en", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 698800, "num_examples": 2160, "dataset_name": "covost2"}, "validation": {"name": "validation", "num_bytes": 438319, "num_examples": 1349, "dataset_name": "covost2"}, "test": {"name": "test", "num_bytes": 517738, "num_examples": 1595, "dataset_name": "covost2"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/covost/covost_v2.sv-SE_en.tsv.tar.gz": {"num_bytes": 96161, "checksum": "f4d9931ef0884eb601e1d6ac7b95893ac0b88534c72c1404c64ae0b4bd2d7c81"}}, "download_size": 96161, "post_processing_size": null, "dataset_size": 1654857, "size_in_bytes": 1751018}, "lv_en": {"description": "\nCoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla\u2019s open source Common Voice database of crowdsourced voice recordings.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .mp3 format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport torchaudio\n\ndef map_to_array(batch):\n    speech_array, _ = torchaudio.load(batch[\"file\"])\n    batch[\"speech\"] = speech_array.numpy()\n    return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "\n@misc{wang2020covost,\n    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},\n    author={Changhan Wang and Anne Wu and Juan Pino},\n    year={2020},\n    eprint={2007.10310},\n    archivePrefix={arXiv},\n    primaryClass={cs.CL}\n", "homepage": "https://github.com/facebookresearch/covost", "license": "", "features": {"client_id": {"dtype": "string", "id": null, "_type": "Value"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "translation": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "translation"}, "builder_name": "covost2", "config_name": "lv_en", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 747290, "num_examples": 2337, "dataset_name": "covost2"}, "validation": {"name": "validation", "num_bytes": 360941, "num_examples": 1125, "dataset_name": "covost2"}, "test": {"name": "test", "num_bytes": 519183, "num_examples": 1629, "dataset_name": "covost2"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/covost/covost_v2.lv_en.tsv.tar.gz": {"num_bytes": 88836, "checksum": "c13ae5b001698f772c683450d121c9a9dccdfd0c90184f75429d65e9f846dfe3"}}, "download_size": 88836, "post_processing_size": null, "dataset_size": 1627414, "size_in_bytes": 1716250}, "sl_en": {"description": "\nCoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla\u2019s open source Common Voice database of crowdsourced voice recordings.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .mp3 format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport torchaudio\n\ndef map_to_array(batch):\n    speech_array, _ = torchaudio.load(batch[\"file\"])\n    batch[\"speech\"] = speech_array.numpy()\n    return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "\n@misc{wang2020covost,\n    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},\n    author={Changhan Wang and Anne Wu and Juan Pino},\n    year={2020},\n    eprint={2007.10310},\n    archivePrefix={arXiv},\n    primaryClass={cs.CL}\n", "homepage": "https://github.com/facebookresearch/covost", "license": "", "features": {"client_id": {"dtype": "string", "id": null, "_type": "Value"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "translation": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "translation"}, "builder_name": "covost2", "config_name": "sl_en", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 602420, "num_examples": 1843, "dataset_name": "covost2"}, "validation": {"name": "validation", "num_bytes": 165977, "num_examples": 509, "dataset_name": "covost2"}, "test": {"name": "test", "num_bytes": 115414, "num_examples": 360, "dataset_name": "covost2"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/covost/covost_v2.sl_en.tsv.tar.gz": {"num_bytes": 58445, "checksum": "6f2e6f2bc8018fff0466030cd385837e89e482d162bee9d5bca269dcbc7eab69"}}, "download_size": 58445, "post_processing_size": null, "dataset_size": 883811, "size_in_bytes": 942256}, "ta_en": {"description": "\nCoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla\u2019s open source Common Voice database of crowdsourced voice recordings.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .mp3 format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport torchaudio\n\ndef map_to_array(batch):\n    speech_array, _ = torchaudio.load(batch[\"file\"])\n    batch[\"speech\"] = speech_array.numpy()\n    return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "\n@misc{wang2020covost,\n    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},\n    author={Changhan Wang and Anne Wu and Juan Pino},\n    year={2020},\n    eprint={2007.10310},\n    archivePrefix={arXiv},\n    primaryClass={cs.CL}\n", "homepage": "https://github.com/facebookresearch/covost", "license": "", "features": {"client_id": {"dtype": "string", "id": null, "_type": "Value"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "translation": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "translation"}, "builder_name": "covost2", "config_name": "ta_en", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 534564, "num_examples": 1358, "dataset_name": "covost2"}, "validation": {"name": "validation", "num_bytes": 150428, "num_examples": 384, "dataset_name": "covost2"}, "test": {"name": "test", "num_bytes": 303843, "num_examples": 786, "dataset_name": "covost2"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/covost/covost_v2.ta_en.tsv.tar.gz": {"num_bytes": 55659, "checksum": "03176367953cde717c1b6c19d21f2a0298f07f3459f5ef24866a44988eaea335"}}, "download_size": 55659, "post_processing_size": null, "dataset_size": 988835, "size_in_bytes": 1044494}, "ja_en": {"description": "\nCoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla\u2019s open source Common Voice database of crowdsourced voice recordings.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .mp3 format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport torchaudio\n\ndef map_to_array(batch):\n    speech_array, _ = torchaudio.load(batch[\"file\"])\n    batch[\"speech\"] = speech_array.numpy()\n    return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "\n@misc{wang2020covost,\n    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},\n    author={Changhan Wang and Anne Wu and Juan Pino},\n    year={2020},\n    eprint={2007.10310},\n    archivePrefix={arXiv},\n    primaryClass={cs.CL}\n", "homepage": "https://github.com/facebookresearch/covost", "license": "", "features": {"client_id": {"dtype": "string", "id": null, "_type": "Value"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "translation": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "translation"}, "builder_name": "covost2", "config_name": "ja_en", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 396334, "num_examples": 1119, "dataset_name": "covost2"}, "validation": {"name": "validation", "num_bytes": 226054, "num_examples": 635, "dataset_name": "covost2"}, "test": {"name": "test", "num_bytes": 241310, "num_examples": 684, "dataset_name": "covost2"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/covost/covost_v2.ja_en.tsv.tar.gz": {"num_bytes": 54666, "checksum": "2ca0ca6607a58cee36c72c43b916a5ffb9bc920a766806751aea6e1620a4371f"}}, "download_size": 54666, "post_processing_size": null, "dataset_size": 863698, "size_in_bytes": 918364}, "id_en": {"description": "\nCoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla\u2019s open source Common Voice database of crowdsourced voice recordings.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .mp3 format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport torchaudio\n\ndef map_to_array(batch):\n    speech_array, _ = torchaudio.load(batch[\"file\"])\n    batch[\"speech\"] = speech_array.numpy()\n    return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "\n@misc{wang2020covost,\n    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},\n    author={Changhan Wang and Anne Wu and Juan Pino},\n    year={2020},\n    eprint={2007.10310},\n    archivePrefix={arXiv},\n    primaryClass={cs.CL}\n", "homepage": "https://github.com/facebookresearch/covost", "license": "", "features": {"client_id": {"dtype": "string", "id": null, "_type": "Value"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "translation": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "translation"}, "builder_name": "covost2", "config_name": "id_en", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 406989, "num_examples": 1243, "dataset_name": "covost2"}, "validation": {"name": "validation", "num_bytes": 259134, "num_examples": 792, "dataset_name": "covost2"}, "test": {"name": "test", "num_bytes": 277053, "num_examples": 844, "dataset_name": "covost2"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/covost/covost_v2.id_en.tsv.tar.gz": {"num_bytes": 51755, "checksum": "4e6b58e66b5a7cbba543895ee1ffbdf5b49018e1274f7fe5b045ecb6be467095"}}, "download_size": 51755, "post_processing_size": null, "dataset_size": 943176, "size_in_bytes": 994931}, "cy_en": {"description": "\nCoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla\u2019s open source Common Voice database of crowdsourced voice recordings.\n\nNote that in order to limit the required storage for preparing this dataset, the audio\nis stored in the .mp3 format and is not converted to a float32 array. To convert, the audio\nfile to a float32 array, please make use of the `.map()` function as follows:\n\n\n```python\nimport torchaudio\n\ndef map_to_array(batch):\n    speech_array, _ = torchaudio.load(batch[\"file\"])\n    batch[\"speech\"] = speech_array.numpy()\n    return batch\n\ndataset = dataset.map(map_to_array, remove_columns=[\"file\"])\n```\n", "citation": "\n@misc{wang2020covost,\n    title={CoVoST 2: A Massively Multilingual Speech-to-Text Translation Corpus},\n    author={Changhan Wang and Anne Wu and Juan Pino},\n    year={2020},\n    eprint={2007.10310},\n    archivePrefix={arXiv},\n    primaryClass={cs.CL}\n", "homepage": "https://github.com/facebookresearch/covost", "license": "", "features": {"client_id": {"dtype": "string", "id": null, "_type": "Value"}, "file": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "translation": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": {"input": "file", "output": "translation"}, "builder_name": "covost2", "config_name": "cy_en", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 432071, "num_examples": 1241, "dataset_name": "covost2"}, "validation": {"name": "validation", "num_bytes": 236107, "num_examples": 690, "dataset_name": "covost2"}, "test": {"name": "test", "num_bytes": 236713, "num_examples": 690, "dataset_name": "covost2"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/covost/covost_v2.cy_en.tsv.tar.gz": {"num_bytes": 875557, "checksum": "73fd49342e27636ab4bd84b3c2762a2803d4ddaa01b98032ba79228335e9560c"}}, "download_size": 875557, "post_processing_size": null, "dataset_size": 904891, "size_in_bytes": 1780448}}

dummy/ar_en/1.0.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ac00b727d99e594ceb969c2558375a2907eee9b3623c2464c17ddcca320a875a
+size 1676

dummy/ca_en/1.0.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:12d7c0f9fcfa0b12d9c6735b41f21052ad02b45abefe414e51e158d1468c9287
+size 1858

dummy/cy_en/1.0.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d0075280642878253c84c701b2868cf6e8a65e20a65790f22ece2a6409ac27db
+size 1719

dummy/de_en/1.0.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e89e26c500384df3a76d5dd13c69fd84f27697b8abfd166ee3b906d6d596488f
+size 1729

dummy/en_ar/1.0.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1722b327049fbd0f6f6bcb20cd247c8476da140554c3fd989c31ceea1a4a1cfe
+size 1795

dummy/en_ca/1.0.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fc00f1653d6a074fc9f3f3a3b8633c8db40c5e867b54753e6020e97084f9edf
+size 1778

dummy/en_cy/1.0.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2a73155aae0d3ddfce253b464304c678b5e6fc71d0d852f3f58a273181fc5e95
+size 1786

dummy/en_de/1.0.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:663b332443195cec14657b0c654c8efcc7d6f5171fca9e7a459d58a3e6cfd739
+size 1784

dummy/en_et/1.0.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c8559c5a5597df067497c365894639621fbe2a9a686d5052d69fd711b09d8ea4
+size 1740

dummy/en_fa/1.0.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0521e9acb62c82bca0fc910e9955ba62e12a9e2635c16db0d6f9891b6e29533e
+size 1822

dummy/en_id/1.0.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:58444e76f94f47392d76ff03c1bd3d3896b723f53d86396db2e28651aa4427a9
+size 1722

dummy/en_ja/1.0.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:150a1d0e08687271f16a3eed33d9b496017a24318d118081d27a55ff040f8611
+size 1853

dummy/en_lv/1.0.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2e532d446353250338505af2f6b8e6a1908b58fe9733f6c876427e0bd803a43a
+size 1778

dummy/en_mn/1.0.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d323255cc5051707c09a4e60230ee5c9546a795cf14a96ad3e419b56d8630776
+size 1874

dummy/en_sl/1.0.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b4e0708afea9618a2731e5fa59673c1fc2422f23717bf7fbb1e707bbb93adc32
+size 1742

dummy/en_sv-SE/1.0.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:028e8ff638fc730ae8e754585d6430b82711a907ffc98de0f3f68bfe934423cd
+size 1779

dummy/en_ta/1.0.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2b6df9136ee4abfc7928b933e4892ad7f04273d2df4488fbf5e40588feb1751c
+size 1905

dummy/en_tr/1.0.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a654d0fe659e39b526a1e47326ce42e168220abec9e9e331b6555a6df3d4e9f5
+size 1769

dummy/en_zh-CN/1.0.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:614f05ce77bafebe4f4da1e881cdd75c38f7451d1498072a391a78f716b8db70
+size 1785

dummy/es_en/1.0.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:29e39949d1310913f2ab2271532b3222371b5f6312932920432263797a69e949
+size 1802

dummy/et_en/1.0.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:adb1594cc882d5d47d82a6e53e1bb658b6c8f33af7df1db708a0839862d53bfe
+size 1827

dummy/fa_en/1.0.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e4b57573f2b7f2607a1d1af4d81bac8e7326868d865a561370f78a7db388804b
+size 1738

dummy/fr_en/1.0.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ac2b163e9a90c12a0a7172ad3565ae96e7fad70e48be41dd854c3aa70dc0a936
+size 1912

dummy/id_en/1.0.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:56b44eca03f92f380a2e3e0404cf14bb6c0de8d99e6d3b726d529703631fc09c
+size 1680

dummy/it_en/1.0.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bde58c3995480d6fa677056a5e9ba727053c4fb5e9283fa27d4608151a6521fe
+size 1846

dummy/ja_en/1.0.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:50219187355f2500a62cba451f75d30ddfca605083f9fea72fe8ebba6d704370
+size 1710

dummy/lv_en/1.0.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:84b057dce4aff8f1cd4c99e4e55efd0da9a1619ecac4f53732e4fc16e5eb9314
+size 1680

dummy/mn_en/1.0.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a988cdc6405d9b4247468a2df2b66a8256899ec79d3d3f331bb33caf2c5bbb95
+size 2002

dummy/nl_en/1.0.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3cfca8d02f2aea257fddb29ccb2eaa778579b77e381625b7128a9638bb824b56
+size 1773

dummy/pt_en/1.0.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ba0b5266f32c701a433decd74c76575ca512ff2d58884970c294fab78921d358
+size 1733

dummy/ru_en/1.0.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:95a6513ae7a368a2bbe31012079f80495e88ff423a3cfc5ba2d4354e6e7d5c79
+size 2133

dummy/sl_en/1.0.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ab718b93eed923fa5c0dda2b8680e90f5d86b15218493d4f2653501ce7419907
+size 1695

dummy/sv-SE_en/1.0.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:08f714cb26f5b13041b6693f4b9c199b0871e3d7f11da3a6bbb93a2e9298513a
+size 1669

dummy/ta_en/1.0.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:73106fcea28eaf99ad24682a9aa1ca23fecbd20182ee6464e3415c9637b3f635
+size 1721

dummy/tr_en/1.0.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d1bebca42846ce76d88404d023474e35f438eab011fa503ad3fd35ca8e4ba411
+size 1736

dummy/zh-CN_en/1.0.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f3e1d7374856a1c6c3f2fe6dea3d2e21391344ef696c587b3ad67e1fe8ea3a5a
+size 1962