model upload

.gitattributes

@@ -14,3 +14,6 @@
 *.pb filter=lfs diff=lfs merge=lfs -text
 *.pt filter=lfs diff=lfs merge=lfs -text
 *.pth filter=lfs diff=lfs merge=lfs -text
+model.ckpt filter=lfs diff=lfs merge=lfs -text
+tokenizer.ckpt filter=lfs diff=lfs merge=lfs -text
+wav2vec2.ckpt filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,115 @@
+---
+language: "en"
+thumbnail:
+tags:
+- ASR
+- CTC
+- Attention
+- Transformers
+- pytorch
+license: "apache-2.0"
+datasets:
+- aishell
+metrics:
+- wer
+- cer
+---
+
+<iframe src="https://ghbtns.com/github-btn.html?user=speechbrain&repo=speechbrain&type=star&count=true&size=large&v=2" frameborder="0" scrolling="0" width="170" height="30" title="GitHub"></iframe>
+<br/><br/>
+
+# Transformer for AISHELL (Mandarin Chinese)
+
+This repository provides all the necessary tools to perform automatic speech
+recognition from an end-to-end system pretrained on AISHELL (Mandarin Chinese)
+within SpeechBrain. For a better experience, we encourage you to learn more about
+[SpeechBrain](https://speechbrain.github.io).
+
+The performance of the model is the following:
+
+| Release | Dev CER | Test CER | GPUs | Full Results |
+|:-------------:|:--------------:|:--------------:|:--------:|:--------:|
+| 05-03-21 | 5.60 | 6.04 | 2xV100 32GB | [Google Drive](https://drive.google.com/drive/folders/1zlTBib0XEwWeyhaXDXnkqtPsIBI18Uzs?usp=sharing)|
+
+
+
+## Pipeline description
+
+This ASR system is composed of 2 different but linked blocks:
+- Tokenizer (unigram) that transforms words into subword units and trained with
+the train transcriptions of LibriSpeech.
+- Acoustic model made of a transformer encoder and a joint decoder with CTC +
+transformer. Hence, the decoding also incorporates the CTC probabilities.
+
+To Train this system from scratch, [see our SpeechBrain recipe](https://github.com/speechbrain/speechbrain/tree/develop/recipes/AISHELL-1).
+
+
+## Install SpeechBrain
+
+First of all, please install SpeechBrain with the following command:
+
+```
+pip install speechbrain
+```
+
+Please notice that we encourage you to read our tutorials and learn more about
+[SpeechBrain](https://speechbrain.github.io).
+
+### Transcribing your own audio files (in English)
+
+```python
+from speechbrain.pretrained import EncoderDecoderASR
+
+asr_model = EncoderDecoderASR.from_hparams(source="speechbrain/asr-transformer-aishell", savedir="pretrained_models/asr-transformer-aishell")
+asr_model.transcribe_file("speechbrain/asr-transformer-aishell/example_mandarin.wav")
+
+```
+
+### Inference on GPU
+To perform inference on the GPU, add  `run_opts={"device":"cuda"}`  when calling the `from_hparams` method.
+
+### Training
+The model was trained with SpeechBrain (Commit hash: '986a2175').
+To train it from scratch follow these steps:
+1. Clone SpeechBrain:
+```bash
+git clone https://github.com/speechbrain/speechbrain/
+```
+2. Install it:
+```bash
+cd speechbrain
+pip install -r requirements.txt
+pip install -e .
+```
+
+3. Run Training:
+```bash
+cd recipes/AISHELL-1/ASR/transformer/
+python train.py hparams/train_ASR_transformer.yaml --data_folder=your_data_folder
+```
+
+You can find our training results (models, logs, etc) [here](https://drive.google.com/drive/folders/1QU18YoauzLOXueogspT0CgR5bqJ6zFfu?usp=sharing).
+
+### Limitations
+The SpeechBrain team does not provide any warranty on the performance achieved by this model when used on other datasets.
+
+
+# **About SpeechBrain**
+- Website: https://speechbrain.github.io/
+- Code: https://github.com/speechbrain/speechbrain/
+- HuggingFace: https://huggingface.co/speechbrain/
+
+
+# **Citing SpeechBrain**
+Please, cite SpeechBrain if you use it for your research or business.
+
+```bibtex
+@misc{speechbrain,
+  title={SpeechBrain: A General-Purpose Speech Toolkit},
+  author={Mirco Ravanelli and Titouan Parcollet and Peter Plantinga and Aku Rouhe and Samuele Cornell and Loren Lugosch and Cem Subakan and Nauman Dawalatabad and Abdelwahab Heba and Jianyuan Zhong and Ju-Chieh Chou and Sung-Lin Yeh and Szu-Wei Fu and Chien-Feng Liao and Elena Rastorgueva and François Grondin and William Aris and Hwidong Na and Yan Gao and Renato De Mori and Yoshua Bengio},
+  year={2021},
+  eprint={2106.04624},
+  archivePrefix={arXiv},
+  primaryClass={eess.AS}
+}
+```

hyperparams.yaml

@@ -0,0 +1,113 @@
+# ############################################################################
+# Model: E2E ASR with Transformer
+# Encoder: Transformer Encoder
+# Decoder: Transformer Decoder + (CTC/ATT joint) beamsearch
+# Tokens: BPE with unigram
+# losses: CTC + KLdiv (Label Smoothing loss)
+# Training: AISHELL-1
+# Authors:  Jianyuan Zhong, Titouan Parcollet
+# ############################################################################
+
+# Feature parameters
+sample_rate: 16000
+n_fft: 400
+n_mels: 80
+wav2vec2_hub: facebook/wav2vec2-large-it-voxpopuli
+
+####################### Model parameters ###########################
+# Transformer
+d_model: 256
+nhead: 4
+num_encoder_layers: 2
+num_decoder_layers: 6
+d_ffn: 2048
+transformer_dropout: 0.1
+activation: !name:torch.nn.GELU
+output_neurons: 5000
+vocab_size: 5000
+
+# Outputs
+blank_index: 0
+label_smoothing: 0.1
+pad_index: 0
+bos_index: 1
+eos_index: 2
+unk_index: 0
+
+# Decoding parameters
+min_decode_ratio: 0.0
+max_decode_ratio: 1.0 # 1.0
+valid_search_interval: 10
+valid_beam_size: 10
+test_beam_size: 10
+ctc_weight_decode: 0.40
+
+############################## models ################################
+
+wav2vec2: !new:speechbrain.lobes.models.huggingface_wav2vec.HuggingFaceWav2Vec2
+    source: !ref <wav2vec2_hub>
+    output_norm: True
+    freeze: True
+    pretrain: False # Pretraining is managed by the SpeechBrain pre-trainer.
+    save_path: !ref <save_folder>/wav2vec2_checkpoint
+
+Transformer: !new:speechbrain.lobes.models.transformer.TransformerASR.TransformerASR # yamllint disable-line rule:line-length
+    input_size: 1024
+    tgt_vocab: !ref <output_neurons>
+    d_model: !ref <d_model>
+    nhead: !ref <nhead>
+    num_encoder_layers: !ref <num_encoder_layers>
+    num_decoder_layers: !ref <num_decoder_layers>
+    d_ffn: !ref <d_ffn>
+    dropout: !ref <transformer_dropout>
+    activation: !ref <activation>
+    normalize_before: True
+
+
+ctc_lin: !new:speechbrain.nnet.linear.Linear
+    input_size: !ref <d_model>
+    n_neurons: !ref <output_neurons>
+
+seq_lin: !new:speechbrain.nnet.linear.Linear
+    input_size: !ref <d_model>
+    n_neurons: !ref <output_neurons>
+
+tokenizer: !new:sentencepiece.SentencePieceProcessor
+
+asr_model: !new:torch.nn.ModuleList
+    - [!ref <Transformer>, !ref <seq_lin>, !ref <ctc_lin>]
+
+# Here, we extract the encoder from the Transformer model
+Tencoder: !new:speechbrain.lobes.models.transformer.TransformerASR.EncoderWrapper
+    transformer: !ref <Transformer>
+
+# We compose the inference (encoder) pipeline.
+encoder: !new:speechbrain.nnet.containers.LengthsCapableSequential
+    input_shape: [null, null, !ref <n_mels>]
+    wav2vec2: !ref <wav2vec2>
+    transformer_encoder: !ref <Tencoder>
+
+decoder: !new:speechbrain.decoders.S2STransformerBeamSearch
+    modules: [!ref <Transformer>, !ref <seq_lin>, !ref <ctc_lin>]
+    bos_index: !ref <bos_index>
+    eos_index: !ref <eos_index>
+    blank_index: !ref <blank_index>
+    min_decode_ratio: !ref <min_decode_ratio>
+    max_decode_ratio: !ref <max_decode_ratio>
+    beam_size: !ref <test_beam_size>
+    ctc_weight: !ref <ctc_weight_decode>
+    using_eos_threshold: False
+    length_normalization: True
+
+modules:
+    encoder: !ref <encoder>
+    decoder: !ref <decoder>
+
+log_softmax: !new:torch.nn.LogSoftmax
+    dim: -1
+
+pretrainer: !new:speechbrain.utils.parameter_transfer.Pretrainer
+    loadables:
+ 	wav2vect2: !ref <wav2vect2>
+        model: !ref <model>
+        tokenizer: !ref <tokenizer>

model.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3edcc685cc45c2775ff90eaad6631e8db3f7de2154e479818a86d3559a6b7bee
+size 67484671

tokenizer.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9f0478ccc6dac61ce0e6149a84e531ff7d300b133d5717cc9d531b00837ac444
+size 300111

wav2vec2.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:98dc840993f8ddd3151611909be73a740d2be8bdf3621437e8ffd738e2a3a6b8
+size 1261930757