nairaxo commited on
Commit
2212563
1 Parent(s): 0403ab7

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +12 -19
README.md CHANGED
@@ -15,17 +15,15 @@ metrics:
15
  - cer
16
  ---
17
 
18
-
19
  <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>
20
  <br/><br/>
21
 
22
- # About
23
- ### DVoice
24
  DVoice is a community initiative that aims to provide Africa low resources languages with data and models to facilitate their use of voice technologies. The lack of data on these languages makes it necessary to collect data using methods that are specific to each one. Two different approaches are currently used: the DVoice platforms ([https://dvoice.ma](https://dvoice.ma) and [https://dvoice.sn](https://dvoice.sn)), which are based on Mozilla Common Voice, for collecting authentic recordings from the community, and transfer learning techniques for automatically labeling recordings that are retrived from social modias. The DVoice platform currently manages 7 languages including Darija (Moroccan Arabic dialect) whose dataset appears on this version, Wolof, Mandingo, Serere, Pular, Diola and Soninke.
25
 
26
  For this project, AIOX Labs the SI2M Laboratory are joining forces to build the future of technologies together.
27
 
28
- ### AIOX Labs
29
  Based in Rabat, London and Paris, AIOX-Labs mobilizes artificial intelligence technologies to meet the business needs and data projects of companies.
30
 
31
  - He is at the service of the growth of groups, the optimization of processes or the improvement of the customer experience.
@@ -35,12 +33,12 @@ Based in Rabat, London and Paris, AIOX-Labs mobilizes artificial intelligence te
35
 
36
  Website: [https://www.aiox-labs.com/](https://www.aiox-labs.com/)
37
 
38
- ### SI2M Laboratory
39
  The Information Systems, Intelligent Systems and Mathematical Modeling Research Laboratory (SI2M) is an academic research laboratory of the National Institute of Statistics and Applied Economics (INSEA). The research areas of the laboratories are Information Systems, Intelligent Systems, Artificial Intelligence, Decision Support, Network and System Security, Mathematical Modelling.
40
 
41
  Website: [SI2M Laboratory](https://insea.ac.ma/index.php/pole-recherche/equipe-de-recherche/150-laboratoire-de-recherche-en-systemes-d-information-systemes-intelligents-et-modelisation-mathematique)
42
 
43
- ### SpeechBrain
44
  SpeechBrain is an open-source and all-in-one speech toolkit. It is designed to be simple, extremely flexible, and user-friendly. Competitive or state-of-the-art performance is obtained in various domains.
45
  Website: https://speechbrain.github.io/
46
  GitHub: https://github.com/speechbrain/speechbrain
@@ -55,7 +53,7 @@ SpeechBrain. For a better experience, we encourage you to learn more about
55
  |:-------------:|:---------------------------:| -----:| -----:| -----:|
56
  | v2.0 | 5.51 | 18.46 | 5.85 | 18.28 |
57
 
58
- ## Pipeline description
59
  This ASR system is composed of 2 different but linked blocks:
60
  - Tokenizer (unigram) that transforms words into subword units and trained with
61
  the train transcriptions.
@@ -64,7 +62,7 @@ The obtained final acoustic representation is given to the CTC greedy decoder.
64
  The system is trained with recordings sampled at 16kHz (single channel).
65
  The code will automatically normalize your audio (i.e., resampling + mono channel selection) when calling *transcribe_file* if needed.
66
 
67
- ## Install SpeechBrain
68
  First of all, please install tranformers and SpeechBrain with the following command:
69
  ```
70
  pip install speechbrain transformers
@@ -72,23 +70,23 @@ pip install speechbrain transformers
72
  Please notice that we encourage you to read the SpeechBrain tutorials and learn more about
73
  [SpeechBrain](https://speechbrain.github.io).
74
 
75
- ### Transcribing your own audio files (in Darija)
76
  ```python
77
  from speechbrain.pretrained import EncoderASR
78
  asr_model = EncoderASR.from_hparams(source="nairaxo/dvoice-darija", savedir="pretrained_models/asr-wav2vec2-dvoice-dar")
79
  asr_model.transcribe_file('./the_path_to_your_audio_file')
80
  ```
81
 
82
- ### Inference on GPU
83
  To perform inference on the GPU, add `run_opts={"device":"cuda"}` when calling the `from_hparams` method.
84
 
85
- ### Training
86
  To train the model from scratch, please see our GitHub tutorial [here](https://github.com/AIOXLABS/DVoice).
87
 
88
- ### Limitations
89
  The SpeechBrain team does not provide any warranty on the performance achieved by this model when used on other datasets.
90
 
91
- #### Referencing SpeechBrain
92
  ```
93
  @misc{SB2021,
94
  author = {Ravanelli, Mirco and Parcollet, Titouan and Rouhe, Aku and Plantinga, Peter and Rastorgueva, Elena and Lugosch, Loren and Dawalatabad, Nauman and Ju-Chieh, Chou and Heba, Abdel and Grondin, Francois and Aris, William and Liao, Chien-Feng and Cornell, Samuele and Yeh, Sung-Lin and Na, Hwidong and Gao, Yan and Fu, Szu-Wei and Subakan, Cem and De Mori, Renato and Bengio, Yoshua },
@@ -98,9 +96,4 @@ The SpeechBrain team does not provide any warranty on the performance achieved b
98
  journal = {GitHub repository},
99
  howpublished = {\\\\url{https://github.com/speechbrain/speechbrain}},
100
  }
101
- ```
102
-
103
- #### About SpeechBrain
104
- SpeechBrain is an open-source and all-in-one speech toolkit. It is designed to be simple, extremely flexible, and user-friendly. Competitive or state-of-the-art performance is obtained in various domains.
105
- Website: https://speechbrain.github.io/
106
- GitHub: https://github.com/speechbrain/speechbrain
 
15
  - cer
16
  ---
17
 
 
18
  <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>
19
  <br/><br/>
20
 
21
+ # DVoice
 
22
  DVoice is a community initiative that aims to provide Africa low resources languages with data and models to facilitate their use of voice technologies. The lack of data on these languages makes it necessary to collect data using methods that are specific to each one. Two different approaches are currently used: the DVoice platforms ([https://dvoice.ma](https://dvoice.ma) and [https://dvoice.sn](https://dvoice.sn)), which are based on Mozilla Common Voice, for collecting authentic recordings from the community, and transfer learning techniques for automatically labeling recordings that are retrived from social modias. The DVoice platform currently manages 7 languages including Darija (Moroccan Arabic dialect) whose dataset appears on this version, Wolof, Mandingo, Serere, Pular, Diola and Soninke.
23
 
24
  For this project, AIOX Labs the SI2M Laboratory are joining forces to build the future of technologies together.
25
 
26
+ # AIOX Labs
27
  Based in Rabat, London and Paris, AIOX-Labs mobilizes artificial intelligence technologies to meet the business needs and data projects of companies.
28
 
29
  - He is at the service of the growth of groups, the optimization of processes or the improvement of the customer experience.
 
33
 
34
  Website: [https://www.aiox-labs.com/](https://www.aiox-labs.com/)
35
 
36
+ # SI2M Laboratory
37
  The Information Systems, Intelligent Systems and Mathematical Modeling Research Laboratory (SI2M) is an academic research laboratory of the National Institute of Statistics and Applied Economics (INSEA). The research areas of the laboratories are Information Systems, Intelligent Systems, Artificial Intelligence, Decision Support, Network and System Security, Mathematical Modelling.
38
 
39
  Website: [SI2M Laboratory](https://insea.ac.ma/index.php/pole-recherche/equipe-de-recherche/150-laboratoire-de-recherche-en-systemes-d-information-systemes-intelligents-et-modelisation-mathematique)
40
 
41
+ # SpeechBrain
42
  SpeechBrain is an open-source and all-in-one speech toolkit. It is designed to be simple, extremely flexible, and user-friendly. Competitive or state-of-the-art performance is obtained in various domains.
43
  Website: https://speechbrain.github.io/
44
  GitHub: https://github.com/speechbrain/speechbrain
 
53
  |:-------------:|:---------------------------:| -----:| -----:| -----:|
54
  | v2.0 | 5.51 | 18.46 | 5.85 | 18.28 |
55
 
56
+ # Pipeline description
57
  This ASR system is composed of 2 different but linked blocks:
58
  - Tokenizer (unigram) that transforms words into subword units and trained with
59
  the train transcriptions.
 
62
  The system is trained with recordings sampled at 16kHz (single channel).
63
  The code will automatically normalize your audio (i.e., resampling + mono channel selection) when calling *transcribe_file* if needed.
64
 
65
+ # Install SpeechBrain
66
  First of all, please install tranformers and SpeechBrain with the following command:
67
  ```
68
  pip install speechbrain transformers
 
70
  Please notice that we encourage you to read the SpeechBrain tutorials and learn more about
71
  [SpeechBrain](https://speechbrain.github.io).
72
 
73
+ # Transcribing your own audio files (in Darija)
74
  ```python
75
  from speechbrain.pretrained import EncoderASR
76
  asr_model = EncoderASR.from_hparams(source="nairaxo/dvoice-darija", savedir="pretrained_models/asr-wav2vec2-dvoice-dar")
77
  asr_model.transcribe_file('./the_path_to_your_audio_file')
78
  ```
79
 
80
+ # Inference on GPU
81
  To perform inference on the GPU, add `run_opts={"device":"cuda"}` when calling the `from_hparams` method.
82
 
83
+ # Training
84
  To train the model from scratch, please see our GitHub tutorial [here](https://github.com/AIOXLABS/DVoice).
85
 
86
+ # Limitations
87
  The SpeechBrain team does not provide any warranty on the performance achieved by this model when used on other datasets.
88
 
89
+ # Referencing SpeechBrain
90
  ```
91
  @misc{SB2021,
92
  author = {Ravanelli, Mirco and Parcollet, Titouan and Rouhe, Aku and Plantinga, Peter and Rastorgueva, Elena and Lugosch, Loren and Dawalatabad, Nauman and Ju-Chieh, Chou and Heba, Abdel and Grondin, Francois and Aris, William and Liao, Chien-Feng and Cornell, Samuele and Yeh, Sung-Lin and Na, Hwidong and Gao, Yan and Fu, Szu-Wei and Subakan, Cem and De Mori, Renato and Bengio, Yoshua },
 
96
  journal = {GitHub repository},
97
  howpublished = {\\\\url{https://github.com/speechbrain/speechbrain}},
98
  }
99
+ ```