|
|
|
|
|
|
|
"""test set""" |
|
|
|
|
|
import csv |
|
import os |
|
import json |
|
|
|
import datasets |
|
from datasets.utils.py_utils import size_str |
|
from tqdm import tqdm |
|
import os |
|
|
|
import datasets |
|
|
|
_CITATION = """\ |
|
@inproceedings{panayotov2015librispeech, |
|
title={Librispeech: an ASR corpus based on public domain audio books}, |
|
author={Panayotov, Vassil and Chen, Guoguo and Povey, Daniel and Khudanpur, Sanjeev}, |
|
booktitle={Acoustics, Speech and Signal Processing (ICASSP), 2015 IEEE International Conference on}, |
|
pages={5206--5210}, |
|
year={2015}, |
|
organization={IEEE} |
|
} |
|
""" |
|
|
|
_DESCRIPTION = """\ |
|
Lorem ipsum |
|
""" |
|
_URL = "https://huggingface.co/datasets/j-krzywdziak/test2" |
|
_AUDIO_URL = "https://huggingface.co/datasets/j-krzywdziak/test2/resolve/main" |
|
_DATA_URL = "https://huggingface.co/datasets/j-krzywdziak/test2/raw/main" |
|
_DL_URLS = { |
|
"clean": { |
|
"of": _AUDIO_URL + "/clean/of/examples.zip", |
|
"on": _AUDIO_URL + "/clean/on/examples.zip", |
|
"example": _DATA_URL + "/clean/example.tsv", |
|
"keyword": _DATA_URL + "/clean/keyword.tsv" |
|
}, |
|
"other": { |
|
"of": _AUDIO_URL + "/other/of/examples.zip", |
|
"on": _AUDIO_URL + "/other/on/examples.zip", |
|
"example": _DATA_URL + "/other/example.tsv", |
|
"keyword": _DATA_URL + "/other/keyword.tsv" |
|
}, |
|
"all": { |
|
"clean.of": _AUDIO_URL + "/clean/of/examples.zip", |
|
"clean.on": _AUDIO_URL + "/clean/on/examples.zip", |
|
"other.of": _AUDIO_URL + "/other/of/examples.zip", |
|
"other.on": _AUDIO_URL + "/other/on/examples.zip", |
|
"clean.example": _DATA_URL + "/clean/example.tsv", |
|
"clean.keyword": _DATA_URL + "/clean/keyword.tsv", |
|
"other.example": _DATA_URL + "/other/example.tsv", |
|
"other.keyword": _DATA_URL + "/other/keyword.tsv" |
|
|
|
}, |
|
} |
|
|
|
|
|
|
|
class TestASR(datasets.GeneratorBasedBuilder): |
|
"""Lorem ipsum.""" |
|
VERSION = "0.0.0" |
|
DEFAULT_CONFIG_NAME = "all" |
|
|
|
BUILDER_CONFIGS = [ |
|
datasets.BuilderConfig(name="clean", description="'Clean' speech."), |
|
datasets.BuilderConfig(name="other", description="'Other', more challenging, speech."), |
|
datasets.BuilderConfig(name="all", description="Combined clean and other dataset."), |
|
] |
|
|
|
def _info(self): |
|
return datasets.DatasetInfo( |
|
description=_DESCRIPTION, |
|
features=datasets.Features( |
|
{ |
|
"path": datasets.Value("string"), |
|
"audio": datasets.Audio(sampling_rate=16_000), |
|
"ngram": datasets.Value("string"), |
|
"type": datasets.Value("string") |
|
} |
|
), |
|
supervised_keys=("file", "text"), |
|
homepage=_URL, |
|
citation=_CITATION |
|
) |
|
|
|
def _split_generators(self, dl_manager): |
|
archive_path = dl_manager.download(_DL_URLS[self.config.name]) |
|
|
|
local_extracted_archive = dl_manager.extract(archive_path) if not dl_manager.is_streaming else {} |
|
if self.config.name == "clean": |
|
of_split = [ |
|
datasets.SplitGenerator( |
|
name="of", |
|
gen_kwargs={ |
|
"local_extracted_archive": local_extracted_archive.get("of"), |
|
"files": dl_manager.iter_archive(archive_path["of"]), |
|
"examples": archive_path["example"], |
|
"keywords": archive_path["keyword"] |
|
}, |
|
) |
|
] |
|
on_split = [ |
|
datasets.SplitGenerator( |
|
name="on", |
|
gen_kwargs={ |
|
"local_extracted_archive": local_extracted_archive.get("on"), |
|
"files": dl_manager.iter_archive(archive_path["on"]), |
|
"examples": archive_path["example"], |
|
"keywords": archive_path["keyword"] |
|
}, |
|
) |
|
] |
|
elif self.config.name == "other": |
|
of_split = [ |
|
datasets.SplitGenerator( |
|
name="of", |
|
gen_kwargs={ |
|
"local_extracted_archive": local_extracted_archive.get("of"), |
|
"files": dl_manager.iter_archive(archive_path["of"]), |
|
"examples": archive_path["example"], |
|
"keywords": archive_path["keyword"] |
|
}, |
|
) |
|
] |
|
on_split = [ |
|
datasets.SplitGenerator( |
|
name="on", |
|
gen_kwargs={ |
|
"local_extracted_archive": local_extracted_archive.get("on"), |
|
"files": dl_manager.iter_archive(archive_path["on"]), |
|
"examples": archive_path["example"], |
|
"keywords": archive_path["keyword"] |
|
}, |
|
) |
|
] |
|
elif self.config.name == "all": |
|
of_split = [ |
|
datasets.SplitGenerator( |
|
name="clean.of", |
|
gen_kwargs={ |
|
"local_extracted_archive": local_extracted_archive.get("clean.of"), |
|
"files": dl_manager.iter_archive(archive_path["clean.of"]), |
|
"examples": archive_path["clean.example"], |
|
"keywords": archive_path["clean.keyword"] |
|
}, |
|
), |
|
datasets.SplitGenerator( |
|
name="other.of", |
|
gen_kwargs={ |
|
"local_extracted_archive": local_extracted_archive.get("other.of"), |
|
"files": dl_manager.iter_archive(archive_path["other.of"]), |
|
"examples": archive_path["other.example"], |
|
"keywords": archive_path["other.keyword"] |
|
} |
|
) |
|
] |
|
on_split = [ |
|
datasets.SplitGenerator( |
|
name="clean.on", |
|
gen_kwargs={ |
|
"local_extracted_archive": local_extracted_archive.get("clean.on"), |
|
"files": dl_manager.iter_archive(archive_path["clean.on"]), |
|
"examples": archive_path["clean.example"], |
|
"keywords": archive_path["clean.keyword"] |
|
}, |
|
), |
|
datasets.SplitGenerator( |
|
name="other.on", |
|
gen_kwargs={ |
|
"local_extracted_archive": local_extracted_archive.get("other.on"), |
|
"files": dl_manager.iter_archive(archive_path["other.on"]), |
|
"examples": archive_path["other.example"], |
|
"keywords": archive_path["other.keyword"] |
|
} |
|
) |
|
] |
|
return on_split + of_split |
|
|
|
def _generate_examples(self, files, local_extracted_archive, examples, keywords): |
|
"""Lorem ipsum.""" |
|
audio_data = {} |
|
transcripts = [] |
|
key = 0 |
|
for path, f in files: |
|
audio_data[path] = f.read() |
|
with open(keywords, encoding="utf-8") as f: |
|
next(f) |
|
for row in f: |
|
r = row.split("\t") |
|
path = 'examples/'+r[0] |
|
ngram = r[1] |
|
transcripts.append({ |
|
"path": path, |
|
"ngram": ngram, |
|
"type": "keyword" |
|
}) |
|
with open(examples, encoding="utf-8") as f2: |
|
for row in f2: |
|
r = row.split("\t") |
|
path = 'examples/'+r[0] |
|
ngram = r[1] |
|
transcripts.append({ |
|
"path": path, |
|
"ngram": ngram, |
|
"type": "example" |
|
}) |
|
if audio_data and len(audio_data) == len(transcripts): |
|
for transcript in transcripts: |
|
audio = {"path": transcript["path"], "bytes": audio_data[transcript["path"]]} |
|
yield key, {"audio": audio, **transcript} |
|
key += 1 |
|
audio_data = {} |
|
transcripts = [] |
|
|
