poonehmousavi commited on
Commit
7ae828f
1 Parent(s): 36bf1d9

Upload README.md

Browse files
Files changed (1) hide show
  1. README.md +122 -0
README.md CHANGED
@@ -1,3 +1,125 @@
1
  ---
 
 
 
 
 
 
 
 
 
2
  license: apache-2.0
 
 
 
 
 
 
3
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
+ language:
3
+ - fr
4
+ thumbnail: null
5
+ tags:
6
+ - automatic-speech-recognition
7
+ - transducer
8
+ - Attention
9
+ - pytorch
10
+ - speechbrain
11
  license: apache-2.0
12
+ datasets:
13
+ - common_voice
14
+ metrics:
15
+ - name: Test WER
16
+ type: wer
17
+ value: ' 17.58'
18
  ---
19
+
20
+ <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>
21
+ <br/><br/>
22
+
23
+ # Transducer trained on CommonVoice 14.0 French (No LM)
24
+ This repository provides all the necessary tools to perform automatic speech
25
+ recognition from an end-to-end system within
26
+ SpeechBrain. For a better experience, we encourage you to learn more about
27
+ [SpeechBrain](https://speechbrain.github.io).
28
+ The performance of the model is the following:
29
+
30
+ | Release | Test CER | Test WER | GPUs |
31
+ |:-------------:|:--------------:|:--------------:| :--------:|
32
+ | 15.08.23 | 7.61 | 17.58 | 1xV100 32GB |
33
+
34
+ ## Credits
35
+ The model is provided by [vitas.ai](https://www.vitas.ai/).
36
+
37
+ ## Pipeline description
38
+ This ASR system is composed of 2 different but linked blocks:
39
+
40
+ - Tokenizer (unigram) that transforms words into subword units and trained with
41
+ the train transcriptions (train.tsv) of CommonVoice (en).
42
+ - Transducers augment CTC by adding an autoregressive predictor and a join network.
43
+
44
+ The system is trained with recordings sampled at 16kHz (single channel).
45
+ The code will automatically normalize your audio (i.e., resampling + mono channel selection) when calling *transcribe_file* if needed.
46
+
47
+ ## Install SpeechBrain
48
+ First of all, please install SpeechBrain with the following command:
49
+
50
+ ```
51
+ pip install speechbrain
52
+ ```
53
+
54
+ Please notice that we encourage you to read our tutorials and learn more about
55
+ [SpeechBrain](https://speechbrain.github.io).
56
+
57
+ ### Transcribing your own audio files (in French)
58
+
59
+ ```python
60
+ from speechbrain.pretrained import EncoderDecoderASR
61
+ asr_model = EncoderDecoderASR.from_hparams(source="speechbrain/speechbrain/asr-transducer-commonvoice-14-fr", savedir="pretrained_models/speechbrain/asr-transducer-commonvoice-14-fr")
62
+ asr_model.transcribe_file("speechbrain/speechbrain/asr-transducer-commonvoice-14-fr/example-fr.wav")
63
+ ```
64
+
65
+ ### Inference on GPU
66
+
67
+ To perform inference on the GPU, add `run_opts={"device":"cuda"}` when calling the `from_hparams` method.
68
+
69
+ ## Parallel Inference on a Batch
70
+
71
+ Please, [see this Colab notebook](https://colab.research.google.com/drive/1hX5ZI9S4jHIjahFCZnhwwQmFoGAi3tmu?usp=sharing) to figure out how to transcribe in parallel a batch of input sentences using a pre-trained model.
72
+
73
+ ### Training
74
+
75
+ The model was trained with SpeechBrain (986a2175).
76
+ To train it from scratch follows these steps:
77
+
78
+ 1. Clone SpeechBrain:
79
+
80
+ ```bash
81
+ git clone https://github.com/speechbrain/speechbrain/
82
+ ```
83
+
84
+ 2. Install it:
85
+
86
+ ```
87
+ cd speechbrain
88
+ pip install -r requirements.txt
89
+ pip install -e .
90
+ ```
91
+
92
+ 3. Run Training:
93
+
94
+ ```
95
+ cd recipes/CommonVoice/ASR/transducer
96
+ python train.py hparams/train_fr.yaml --data_folder=your_data_folder
97
+ ```
98
+
99
+ You can find our training results (models, logs, etc) [here](https://www.dropbox.com/sh/nv2pnpo5n3besn3/AADZ7l41oLt11ZuOE4MqoJhCa?dl=0)
100
+
101
+ ### Limitations
102
+
103
+ The SpeechBrain team does not provide any warranty on the performance achieved by this model when used on other datasets.
104
+
105
+ # **About SpeechBrain**
106
+
107
+ - Website: https://speechbrain.github.io/
108
+ - Code: https://github.com/speechbrain/speechbrain/
109
+ - HuggingFace: https://huggingface.co/speechbrain/
110
+
111
+ # **Citing SpeechBrain**
112
+
113
+ Please, cite SpeechBrain if you use it for your research or business.
114
+
115
+ ```bibtex
116
+ @misc{speechbrain,
117
+ title={{SpeechBrain}: A General-Purpose Speech Toolkit},
118
+ 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},
119
+ year={2021},
120
+ eprint={2106.04624},
121
+ archivePrefix={arXiv},
122
+ primaryClass={eess.AS},
123
+ note={arXiv:2106.04624}
124
+ }
125
+ ```