jonatasgrosman commited on
Commit
edb5bd2
1 Parent(s): f251e2b

update model

Browse files
Files changed (2) hide show
  1. README.md +15 -13
  2. pytorch_model.bin +1 -1
README.md CHANGED
@@ -2,6 +2,8 @@
2
  language: en
3
  datasets:
4
  - common_voice
 
 
5
  metrics:
6
  - wer
7
  - cer
@@ -24,15 +26,15 @@ model-index:
24
  metrics:
25
  - name: Test WER
26
  type: wer
27
- value: 19.18
28
  - name: Test CER
29
  type: cer
30
- value: 8.25
31
  ---
32
 
33
  # Wav2Vec2-Large-XLSR-53-English
34
 
35
- Fine-tuned [facebook/wav2vec2-large-xlsr-53](https://huggingface.co/facebook/wav2vec2-large-xlsr-53) on English using the [Common Voice](https://huggingface.co/datasets/common_voice).
36
  When using this model, make sure that your speech input is sampled at 16kHz.
37
 
38
  The script used for training can be found here: https://github.com/jonatasgrosman/wav2vec2-sprint
@@ -81,16 +83,16 @@ for i, predicted_sentence in enumerate(predicted_sentences):
81
 
82
  | Reference | Prediction |
83
  | ------------- | ------------- |
84
- | "SHE'LL BE ALL RIGHT." | SHE'LD BE ALL RIGHT |
85
  | SIX | SIX |
86
  | "ALL'S WELL THAT ENDS WELL." | ALL IS WELL THAT ENDS WELL |
87
  | DO YOU MEAN IT? | DO YOU MEAN IT |
88
  | THE NEW PATCH IS LESS INVASIVE THAN THE OLD ONE, BUT STILL CAUSES REGRESSIONS. | THE NEW PATCH IS LESS INVASIVE THAN THE OLD ONE BUT STILL CAUSES REGRESSION |
89
- | HOW IS MOZILLA GOING TO HANDLE AMBIGUITIES LIKE QUEUE AND CUE? | HOWIS MOCILE ARE GOING TO HANDLE AMBIGUITIES LIKE KU AND KU |
90
- | "I GUESS YOU MUST THINK I'M KINDA BATTY." | RISSHON WAS INCAN IN THE BAK TE |
91
  | NO ONE NEAR THE REMOTE MACHINE YOU COULD RING? | NO ONE NEAR THE REMOTE MACHINE YOU COULD RING |
92
- | SAUCE FOR THE GOOSE IS SAUCE FOR THE GANDER. | SAUCE FOR THE GUISE IS SAUCED FOR THE GONDER |
93
- | GROVES STARTED WRITING SONGS WHEN SHE WAS FOUR YEARS OLD. | GRAFS STARTED WRITING SONGS WHEN SHE WAS FOUR YEARS OLD |
94
 
95
  ## Evaluation
96
 
@@ -164,7 +166,7 @@ print(f"CER: {cer.compute(predictions=predictions, references=references, chunk_
164
 
165
  **Test Result**:
166
 
167
- In the table below I report the Word Error Rate (WER) and the Character Error Rate (CER) of the model. I ran the evaluation script described above on other models as well (on 2021-05-20). Note that the table below may show different results from those already reported, this may have been caused due to some specificity of the other evaluation scripts used.
168
 
169
  ---
170
 
@@ -172,7 +174,7 @@ In the table below I report the Word Error Rate (WER) and the Character Error Ra
172
 
173
  | Model | WER | CER |
174
  | ------------- | ------------- | ------------- |
175
- | jonatasgrosman/wav2vec2-large-xlsr-53-english | **19.18%** | **8.25%** |
176
  | jonatasgrosman/wav2vec2-large-english | 21.16% | 9.53% |
177
  | facebook/wav2vec2-large-960h-lv60-self | 22.03% | 10.39% |
178
  | facebook/wav2vec2-large-960h-lv60 | 23.97% | 11.14% |
@@ -194,8 +196,8 @@ In the table below I report the Word Error Rate (WER) and the Character Error Ra
194
  | facebook/wav2vec2-large-960h-lv60 | 2.15% | 0.61% |
195
  | facebook/wav2vec2-large-960h | 2.82% | 0.84% |
196
  | facebook/wav2vec2-base-960h | 3.44% | 1.06% |
 
197
  | facebook/wav2vec2-base-100h | 6.26% | 2.00% |
198
- | jonatasgrosman/wav2vec2-large-xlsr-53-english | 6.97% | 2.02% |
199
  | jonatasgrosman/wav2vec2-large-english | 8.00% | 2.55% |
200
  | elgeish/wav2vec2-large-lv60-timit-asr | 15.53% | 4.93% |
201
  | boris/xlsr-en-punctuation | 19.28% | 6.45% |
@@ -211,8 +213,8 @@ In the table below I report the Word Error Rate (WER) and the Character Error Ra
211
  | facebook/wav2vec2-large-960h-lv60-self | **3.89%** | **1.40%** |
212
  | facebook/wav2vec2-large-960h-lv60 | 4.45% | 1.56% |
213
  | facebook/wav2vec2-large-960h | 6.49% | 2.52% |
 
214
  | facebook/wav2vec2-base-960h | 8.90% | 3.55% |
215
- | jonatasgrosman/wav2vec2-large-xlsr-53-english | 11.75% | 4.23% |
216
  | jonatasgrosman/wav2vec2-large-english | 13.62% | 5.24% |
217
  | facebook/wav2vec2-base-100h | 13.97% | 5.51% |
218
  | boris/xlsr-en-punctuation | 26.40% | 10.11% |
@@ -228,9 +230,9 @@ In the table below I report the Word Error Rate (WER) and the Character Error Ra
228
  | ------------- | ------------- | ------------- |
229
  | facebook/wav2vec2-large-960h-lv60-self | **5.17%** | **1.33%** |
230
  | facebook/wav2vec2-large-960h-lv60 | 6.24% | 1.54% |
 
231
  | facebook/wav2vec2-large-960h | 9.63% | 2.19% |
232
  | facebook/wav2vec2-base-960h | 11.48% | 2.76% |
233
- | jonatasgrosman/wav2vec2-large-xlsr-53-english | 11.93% | 3.50% |
234
  | elgeish/wav2vec2-large-lv60-timit-asr | 13.83% | 4.36% |
235
  | jonatasgrosman/wav2vec2-large-english | 13.91% | 4.01% |
236
  | facebook/wav2vec2-base-100h | 16.75% | 4.79% |
 
2
  language: en
3
  datasets:
4
  - common_voice
5
+ - librispeech_asr
6
+ - timit_asr
7
  metrics:
8
  - wer
9
  - cer
 
26
  metrics:
27
  - name: Test WER
28
  type: wer
29
+ value: 19.76
30
  - name: Test CER
31
  type: cer
32
+ value: 8.60
33
  ---
34
 
35
  # Wav2Vec2-Large-XLSR-53-English
36
 
37
+ Fine-tuned [facebook/wav2vec2-large-xlsr-53](https://huggingface.co/facebook/wav2vec2-large-xlsr-53) on English using the [Common Voice](https://huggingface.co/datasets/common_voice), [LibriSpeech](https://huggingface.co/datasets/librispeech_asr) and [TIMIT](https://huggingface.co/datasets/timit_asr),.
38
  When using this model, make sure that your speech input is sampled at 16kHz.
39
 
40
  The script used for training can be found here: https://github.com/jonatasgrosman/wav2vec2-sprint
 
83
 
84
  | Reference | Prediction |
85
  | ------------- | ------------- |
86
+ | "SHE'LL BE ALL RIGHT." | SHE'D BE ALRIGHT |
87
  | SIX | SIX |
88
  | "ALL'S WELL THAT ENDS WELL." | ALL IS WELL THAT ENDS WELL |
89
  | DO YOU MEAN IT? | DO YOU MEAN IT |
90
  | THE NEW PATCH IS LESS INVASIVE THAN THE OLD ONE, BUT STILL CAUSES REGRESSIONS. | THE NEW PATCH IS LESS INVASIVE THAN THE OLD ONE BUT STILL CAUSES REGRESSION |
91
+ | HOW IS MOZILLA GOING TO HANDLE AMBIGUITIES LIKE QUEUE AND CUE? | HOW IS MUSILA GOING TO HANDLE ANB HOOTIES LIKE QU AND QU |
92
+ | "I GUESS YOU MUST THINK I'M KINDA BATTY." | RISIONAS INCI IN TE BACTY |
93
  | NO ONE NEAR THE REMOTE MACHINE YOU COULD RING? | NO ONE NEAR THE REMOTE MACHINE YOU COULD RING |
94
+ | SAUCE FOR THE GOOSE IS SAUCE FOR THE GANDER. | SAUCE FOR THE GUISE IS SAUCE FOR THE GONDER |
95
+ | GROVES STARTED WRITING SONGS WHEN SHE WAS FOUR YEARS OLD. | GRAFS STARTED WRITING SOUNDS WHEN SHE WAS FOUR YEARS OLD |
96
 
97
  ## Evaluation
98
 
 
166
 
167
  **Test Result**:
168
 
169
+ In the table below I report the Word Error Rate (WER) and the Character Error Rate (CER) of the model. I ran the evaluation script described above on other models as well (on 2021-05-20). Note that the table below may show different results from those already reported, this may have been caused due to some specificity of the other evaluation scripts used... I've also tested the model using the LibriSpeech and TIMIT datasets, which are better-behaved datasets than the Common Voice, containing only examples in US English extracted from audiobooks.
170
 
171
  ---
172
 
 
174
 
175
  | Model | WER | CER |
176
  | ------------- | ------------- | ------------- |
177
+ | jonatasgrosman/wav2vec2-large-xlsr-53-english | **19.76%** | **8.60%** |
178
  | jonatasgrosman/wav2vec2-large-english | 21.16% | 9.53% |
179
  | facebook/wav2vec2-large-960h-lv60-self | 22.03% | 10.39% |
180
  | facebook/wav2vec2-large-960h-lv60 | 23.97% | 11.14% |
 
196
  | facebook/wav2vec2-large-960h-lv60 | 2.15% | 0.61% |
197
  | facebook/wav2vec2-large-960h | 2.82% | 0.84% |
198
  | facebook/wav2vec2-base-960h | 3.44% | 1.06% |
199
+ | jonatasgrosman/wav2vec2-large-xlsr-53-english | 4.16% | 1.28% |
200
  | facebook/wav2vec2-base-100h | 6.26% | 2.00% |
 
201
  | jonatasgrosman/wav2vec2-large-english | 8.00% | 2.55% |
202
  | elgeish/wav2vec2-large-lv60-timit-asr | 15.53% | 4.93% |
203
  | boris/xlsr-en-punctuation | 19.28% | 6.45% |
 
213
  | facebook/wav2vec2-large-960h-lv60-self | **3.89%** | **1.40%** |
214
  | facebook/wav2vec2-large-960h-lv60 | 4.45% | 1.56% |
215
  | facebook/wav2vec2-large-960h | 6.49% | 2.52% |
216
+ | jonatasgrosman/wav2vec2-large-xlsr-53-english | 8.82% | 3.42% |
217
  | facebook/wav2vec2-base-960h | 8.90% | 3.55% |
 
218
  | jonatasgrosman/wav2vec2-large-english | 13.62% | 5.24% |
219
  | facebook/wav2vec2-base-100h | 13.97% | 5.51% |
220
  | boris/xlsr-en-punctuation | 26.40% | 10.11% |
 
230
  | ------------- | ------------- | ------------- |
231
  | facebook/wav2vec2-large-960h-lv60-self | **5.17%** | **1.33%** |
232
  | facebook/wav2vec2-large-960h-lv60 | 6.24% | 1.54% |
233
+ | jonatasgrosman/wav2vec2-large-xlsr-53-english | 6.81% | 2.02% |
234
  | facebook/wav2vec2-large-960h | 9.63% | 2.19% |
235
  | facebook/wav2vec2-base-960h | 11.48% | 2.76% |
 
236
  | elgeish/wav2vec2-large-lv60-timit-asr | 13.83% | 4.36% |
237
  | jonatasgrosman/wav2vec2-large-english | 13.91% | 4.01% |
238
  | facebook/wav2vec2-base-100h | 16.75% | 4.79% |
pytorch_model.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:3ba6ad16f6ecfcadd07ca5bc3353c3168665bee9fbfb160fbc864a6a9f87ca58
3
  size 1262069143
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f9a3cc8402adf407944f67025114dd0f02af9584bfb266c031f586d601719b0a
3
  size 1262069143