nithinraok commited on
Commit
74ec759
0 Parent(s):

adding parakeet tdt model

Browse files
Files changed (3) hide show
  1. .gitattributes +36 -0
  2. README.md +301 -0
  3. parakeet-tdt-1.1b.nemo +3 -0
.gitattributes ADDED
@@ -0,0 +1,36 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ *.7z filter=lfs diff=lfs merge=lfs -text
2
+ *.arrow filter=lfs diff=lfs merge=lfs -text
3
+ *.bin filter=lfs diff=lfs merge=lfs -text
4
+ *.bz2 filter=lfs diff=lfs merge=lfs -text
5
+ *.ckpt filter=lfs diff=lfs merge=lfs -text
6
+ *.ftz filter=lfs diff=lfs merge=lfs -text
7
+ *.gz filter=lfs diff=lfs merge=lfs -text
8
+ *.h5 filter=lfs diff=lfs merge=lfs -text
9
+ *.joblib filter=lfs diff=lfs merge=lfs -text
10
+ *.lfs.* filter=lfs diff=lfs merge=lfs -text
11
+ *.mlmodel filter=lfs diff=lfs merge=lfs -text
12
+ *.model filter=lfs diff=lfs merge=lfs -text
13
+ *.msgpack filter=lfs diff=lfs merge=lfs -text
14
+ *.npy filter=lfs diff=lfs merge=lfs -text
15
+ *.npz filter=lfs diff=lfs merge=lfs -text
16
+ *.onnx filter=lfs diff=lfs merge=lfs -text
17
+ *.ot filter=lfs diff=lfs merge=lfs -text
18
+ *.parquet filter=lfs diff=lfs merge=lfs -text
19
+ *.pb filter=lfs diff=lfs merge=lfs -text
20
+ *.pickle filter=lfs diff=lfs merge=lfs -text
21
+ *.pkl filter=lfs diff=lfs merge=lfs -text
22
+ *.pt filter=lfs diff=lfs merge=lfs -text
23
+ *.pth filter=lfs diff=lfs merge=lfs -text
24
+ *.rar filter=lfs diff=lfs merge=lfs -text
25
+ *.safetensors filter=lfs diff=lfs merge=lfs -text
26
+ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
27
+ *.tar.* filter=lfs diff=lfs merge=lfs -text
28
+ *.tar filter=lfs diff=lfs merge=lfs -text
29
+ *.tflite filter=lfs diff=lfs merge=lfs -text
30
+ *.tgz filter=lfs diff=lfs merge=lfs -text
31
+ *.wasm filter=lfs diff=lfs merge=lfs -text
32
+ *.xz filter=lfs diff=lfs merge=lfs -text
33
+ *.zip filter=lfs diff=lfs merge=lfs -text
34
+ *.zst filter=lfs diff=lfs merge=lfs -text
35
+ *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ parakeet-tdt-1.1b.nemo filter=lfs diff=lfs merge=lfs -text
README.md ADDED
@@ -0,0 +1,301 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ language:
3
+ - en
4
+ library_name: nemo
5
+ datasets:
6
+ - librispeech_asr
7
+ - fisher_corpus
8
+ - Switchboard-1
9
+ - WSJ-0
10
+ - WSJ-1
11
+ - National-Singapore-Corpus-Part-1
12
+ - National-Singapore-Corpus-Part-6
13
+ - vctk
14
+ - voxpopuli
15
+ - europarl
16
+ - multilingual_librispeech
17
+ - mozilla-foundation/common_voice_8_0
18
+ - MLCommons/peoples_speech
19
+ thumbnail: null
20
+ tags:
21
+ - automatic-speech-recognition
22
+ - speech
23
+ - audio
24
+ - Transducer
25
+ - TDT
26
+ - FastConformer
27
+ - Conformer
28
+ - pytorch
29
+ - NeMo
30
+ - hf-asr-leaderboard
31
+ license: cc-by-4.0
32
+ widget:
33
+ - example_title: Librispeech sample 1
34
+ src: https://cdn-media.huggingface.co/speech_samples/sample1.flac
35
+ - example_title: Librispeech sample 2
36
+ src: https://cdn-media.huggingface.co/speech_samples/sample2.flac
37
+ model-index:
38
+ - name: parakeet_tdt_1.1b
39
+ results:
40
+ - task:
41
+ name: Automatic Speech Recognition
42
+ type: automatic-speech-recognition
43
+ dataset:
44
+ name: AMI (Meetings test)
45
+ type: edinburghcstr/ami
46
+ config: ihm
47
+ split: test
48
+ args:
49
+ language: en
50
+ metrics:
51
+ - name: Test WER
52
+ type: wer
53
+ value: 15.90
54
+ - task:
55
+ name: Automatic Speech Recognition
56
+ type: automatic-speech-recognition
57
+ dataset:
58
+ name: Earnings-22
59
+ type: revdotcom/earnings22
60
+ split: test
61
+ args:
62
+ language: en
63
+ metrics:
64
+ - name: Test WER
65
+ type: wer
66
+ value: 14.65
67
+ - task:
68
+ name: Automatic Speech Recognition
69
+ type: automatic-speech-recognition
70
+ dataset:
71
+ name: GigaSpeech
72
+ type: speechcolab/gigaspeech
73
+ split: test
74
+ args:
75
+ language: en
76
+ metrics:
77
+ - name: Test WER
78
+ type: wer
79
+ value: 9.55
80
+ - task:
81
+ name: Automatic Speech Recognition
82
+ type: automatic-speech-recognition
83
+ dataset:
84
+ name: LibriSpeech (clean)
85
+ type: librispeech_asr
86
+ config: other
87
+ split: test
88
+ args:
89
+ language: en
90
+ metrics:
91
+ - name: Test WER
92
+ type: wer
93
+ value: 1.39
94
+ - task:
95
+ name: Automatic Speech Recognition
96
+ type: automatic-speech-recognition
97
+ dataset:
98
+ name: LibriSpeech (other)
99
+ type: librispeech_asr
100
+ config: other
101
+ split: test
102
+ args:
103
+ language: en
104
+ metrics:
105
+ - name: Test WER
106
+ type: wer
107
+ value: 2.62
108
+ - task:
109
+ type: Automatic Speech Recognition
110
+ name: automatic-speech-recognition
111
+ dataset:
112
+ name: SPGI Speech
113
+ type: kensho/spgispeech
114
+ config: test
115
+ split: test
116
+ args:
117
+ language: en
118
+ metrics:
119
+ - name: Test WER
120
+ type: wer
121
+ value: 3.42
122
+ - task:
123
+ type: Automatic Speech Recognition
124
+ name: automatic-speech-recognition
125
+ dataset:
126
+ name: tedlium-v3
127
+ type: LIUM/tedlium
128
+ config: release1
129
+ split: test
130
+ args:
131
+ language: en
132
+ metrics:
133
+ - name: Test WER
134
+ type: wer
135
+ value: 3.56
136
+ - task:
137
+ name: Automatic Speech Recognition
138
+ type: automatic-speech-recognition
139
+ dataset:
140
+ name: Vox Populi
141
+ type: facebook/voxpopuli
142
+ config: en
143
+ split: test
144
+ args:
145
+ language: en
146
+ metrics:
147
+ - name: Test WER
148
+ type: wer
149
+ value: 5.48
150
+ - task:
151
+ type: Automatic Speech Recognition
152
+ name: automatic-speech-recognition
153
+ dataset:
154
+ name: Mozilla Common Voice 9.0
155
+ type: mozilla-foundation/common_voice_9_0
156
+ config: en
157
+ split: test
158
+ args:
159
+ language: en
160
+ metrics:
161
+ - name: Test WER
162
+ type: wer
163
+ value: 5.97
164
+
165
+ metrics:
166
+ - wer
167
+ pipeline_tag: automatic-speech-recognition
168
+ ---
169
+
170
+ # Parakeet TDT 1.1B (en)
171
+
172
+ <style>
173
+ img {
174
+ display: inline;
175
+ }
176
+ </style>
177
+
178
+ [![Model architecture](https://img.shields.io/badge/Model_Arch-FastConformer--TDT-lightgrey#model-badge)](#model-architecture)
179
+ | [![Model size](https://img.shields.io/badge/Params-1.1B-lightgrey#model-badge)](#model-architecture)
180
+ | [![Language](https://img.shields.io/badge/Language-en-lightgrey#model-badge)](#datasets)
181
+
182
+
183
+ `parakeet-tdt-1.1b` is an ASR model that transcribes speech in lower case English alphabet. This model is jointly developed by [NVIDIA NeMo](https://github.com/NVIDIA/NeMo) and [Suno.ai](https://www.suno.ai/) teams.
184
+ It is an XXL version of FastConformer [1] TDT [2] (around 1.1B parameters) model.
185
+ See the [model architecture](#model-architecture) section and [NeMo documentation](https://docs.nvidia.com/deeplearning/nemo/user-guide/docs/en/main/asr/models.html#fast-conformer) for complete architecture details.
186
+
187
+ ## NVIDIA NeMo: Training
188
+
189
+ To train, fine-tune or play with the model you will need to install [NVIDIA NeMo](https://github.com/NVIDIA/NeMo). We recommend you install it after you've installed latest PyTorch version.
190
+ ```
191
+ pip install nemo_toolkit['all']
192
+ ```
193
+
194
+ ## How to Use this Model
195
+
196
+ The model is available for use in the NeMo toolkit [3], and can be used as a pre-trained checkpoint for inference or for fine-tuning on another dataset.
197
+
198
+ ### Automatically instantiate the model
199
+
200
+ ```python
201
+ import nemo.collections.asr as nemo_asr
202
+ asr_model = nemo_asr.models.EncDecRNNTBPEModel.from_pretrained(model_name="nvidia/parakeet-tdt-1.1b")
203
+ ```
204
+
205
+ ### Transcribing using Python
206
+ First, let's get a sample
207
+ ```
208
+ wget https://dldata-public.s3.us-east-2.amazonaws.com/2086-149220-0033.wav
209
+ ```
210
+ Then simply do:
211
+ ```
212
+ asr_model.transcribe(['2086-149220-0033.wav'])
213
+ ```
214
+
215
+ ### Transcribing many audio files
216
+
217
+ ```shell
218
+ python [NEMO_GIT_FOLDER]/examples/asr/transcribe_speech.py
219
+ pretrained_name="nvidia/parakeet-tdt-1.1b"
220
+ audio_dir="<DIRECTORY CONTAINING AUDIO FILES>"
221
+ ```
222
+
223
+ ### Input
224
+
225
+ This model accepts 16000 Hz mono-channel audio (wav files) as input.
226
+
227
+ ### Output
228
+
229
+ This model provides transcribed speech as a string for a given audio sample.
230
+
231
+ ## Model Architecture
232
+
233
+ This model uses a FastConformer-TDT architecture. FastConformer [1] is an optimized version of the Conformer model with 8x depthwise-separable convolutional downsampling. You may find more information on the details of FastConformer here: [Fast-Conformer Model](https://docs.nvidia.com/deeplearning/nemo/user-guide/docs/en/main/asr/models.html#fast-conformer).
234
+
235
+ TDT (Token-and-Duration Transducer) [2] is a generalization of conventional Transducers by decoupling token and duration predictions. Unlike conventional Transducers which produces a lot of blanks during inference, a TDT model can skip majority of blank predictions by using the duration output (up to 4 frames for this parakeet-tdt-1.1b model), thus brings significant inference speed-up. The detail of TDT can be found here: [Efficient Sequence Transduction by Jointly Predicting Tokens and Durations](https://arxiv.org/abs/2304.06795).
236
+
237
+ ## Training
238
+
239
+ The NeMo toolkit [3] was used for training the models for over several hundred epochs. These model are trained with this [example script](https://github.com/NVIDIA/NeMo/blob/main/examples/asr/asr_transducer/speech_to_text_rnnt_bpe.py) and this [base config](https://github.com/NVIDIA/NeMo/blob/main/examples/asr/conf/fastconformer/fast-conformer_transducer_bpe.yaml).
240
+
241
+ The tokenizers for these models were built using the text transcripts of the train set with this [script](https://github.com/NVIDIA/NeMo/blob/main/scripts/tokenizers/process_asr_text_tokenizer.py).
242
+
243
+ ### Datasets
244
+
245
+ The model was trained on 64K hours of English speech collected and prepared by NVIDIA NeMo and Suno teams.
246
+
247
+ The training dataset consists of private subset with 40K hours of English speech plus 24K hours from the following public datasets:
248
+
249
+ - Librispeech 960 hours of English speech
250
+ - Fisher Corpus
251
+ - Switchboard-1 Dataset
252
+ - WSJ-0 and WSJ-1
253
+ - National Speech Corpus (Part 1, Part 6)
254
+ - VCTK
255
+ - VoxPopuli (EN)
256
+ - Europarl-ASR (EN)
257
+ - Multilingual Librispeech (MLS EN) - 2,000 hour subset
258
+ - Mozilla Common Voice (v7.0)
259
+ - People's Speech - 12,000 hour subset
260
+
261
+ ## Performance
262
+
263
+ The performance of Automatic Speech Recognition models is measuring using Word Error Rate. Since this dataset is trained on multiple domains and a much larger corpus, it will generally perform better at transcribing audio in general.
264
+
265
+ The following tables summarizes the performance of the available models in this collection with the Transducer decoder. Performances of the ASR models are reported in terms of Word Error Rate (WER%) with greedy decoding.
266
+
267
+ |**Version**|**Tokenizer**|**Vocabulary Size**|**AMI**|**Earnings-22**|**Giga Speech**|**LS test-clean**|**SPGI Speech**|**TEDLIUM-v3**|**Vox Populi**|**Common Voice**|
268
+ |---------|-----------------------|-----------------|---------------|---------------|------------|-----------|-----|-------|------|------|
269
+ | 1.22.0 | SentencePiece Unigram | 1024 | 15.90 | 14.65 | 9.55 | 1.39 | 2.62 | 3.42 | 3.56 | 5.48 | 5.97 |
270
+
271
+ These are greedy WER numbers without external LM. More details on evaluation can be found at [HuggingFace ASR Leaderboard](https://huggingface.co/spaces/hf-audio/open_asr_leaderboard)
272
+
273
+ ## NVIDIA Riva: Deployment
274
+
275
+ [NVIDIA Riva](https://developer.nvidia.com/riva), is an accelerated speech AI SDK deployable on-prem, in all clouds, multi-cloud, hybrid, on edge, and embedded.
276
+ Additionally, Riva provides:
277
+
278
+ * World-class out-of-the-box accuracy for the most common languages with model checkpoints trained on proprietary data with hundreds of thousands of GPU-compute hours
279
+ * Best in class accuracy with run-time word boosting (e.g., brand and product names) and customization of acoustic model, language model, and inverse text normalization
280
+ * Streaming speech recognition, Kubernetes compatible scaling, and enterprise-grade support
281
+
282
+ Although this model isn’t supported yet by Riva, the [list of supported models is here](https://huggingface.co/models?other=Riva).
283
+ Check out [Riva live demo](https://developer.nvidia.com/riva#demos).
284
+
285
+ ## References
286
+ [1] [Fast Conformer with Linearly Scalable Attention for Efficient Speech Recognition](https://arxiv.org/abs/2305.05084)
287
+
288
+ [2] [Efficient Sequence Transduction by Jointly Predicting Tokens and Durations](https://arxiv.org/abs/2304.06795)
289
+
290
+ [3] [Google Sentencepiece Tokenizer](https://github.com/google/sentencepiece)
291
+
292
+ [4] [NVIDIA NeMo Toolkit](https://github.com/NVIDIA/NeMo)
293
+
294
+ [5] [Suno.ai](https://suno.ai/)
295
+
296
+ [6] [HuggingFace ASR Leaderboard](https://huggingface.co/spaces/hf-audio/open_asr_leaderboard)
297
+
298
+
299
+ ## Licence
300
+
301
+ License to use this model is covered by the [CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/). By downloading the public and release version of the model, you accept the terms and conditions of the [CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/) license.
parakeet-tdt-1.1b.nemo ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9c563d52bdffeacbac0c5b894fdea9be82fea3a6bd8bb8018ff57888e2b5d988
3
+ size 4283136000