msekoyan commited on
Commit
5e93b07
·
1 Parent(s): cc46424

update readme and add plots

Browse files

Signed-off-by: monica-sekoyan <[email protected]>

Files changed (5) hide show
  1. .gitattributes +1 -0
  2. README.md +9 -9
  3. plots/asr.png +3 -0
  4. plots/en_x.png +3 -0
  5. plots/x_en.png +3 -0
.gitattributes CHANGED
@@ -34,3 +34,4 @@ saved_model/**/* 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
  canary-1b-v2.nemo 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
  canary-1b-v2.nemo filter=lfs diff=lfs merge=lfs -text
37
+ *.png filter=lfs diff=lfs merge=lfs -text
README.md CHANGED
@@ -42,7 +42,7 @@ We will soon release a comprehensive **Canary-1b-v2 technical report** detailing
42
 
43
  ### Automatic Speech Recognition (ASR)
44
 
45
- ![ASR WER Comparison](plots/asr_performance.png)
46
 
47
  *Figure 1: ASR WER comparison across different models. This does not include Punctuation and Capitalisation errors.*
48
 
@@ -52,14 +52,14 @@ We will soon release a comprehensive **Canary-1b-v2 technical report** detailing
52
 
53
  #### X → English
54
 
55
- ![AST X-En Comparison](plots/x_en_performance.png)
56
 
57
  *Figure 2: AST X → En COMET scores comparison across different models*
58
 
59
  #### English → X
60
 
61
 
62
- ![AST En-X Comparison](plots/en_x_performance.png)
63
 
64
  *Figure 3: AST En → X COMET scores comparison across different models*
65
 
@@ -283,7 +283,7 @@ This section reports the evaluation results of the ``Canary-1b-v2`` model across
283
 
284
  | **WER ↓** | Fleurs-25 Langs | CoVoST-13 Langs | MLS - 6 Langs |
285
  | --------------- | -------------------- | -------------------- | ------------------ |
286
- | **`Canary-1b-v2`** | 8.42% | 7.61% | 7.29% |
287
 
288
 
289
  **Note:** Presented WERs do not include Punctuation and Capitalization errors.
@@ -326,9 +326,9 @@ More details on evaluation can be found at [HuggingFace ASR Leaderboard](https:/
326
  Performance across different Signal-to-Noise Ratios (SNR) using MUSAN music and noise samples \[16] on the [LibriSpeech Clean test set](https://www.openslr.org/12).
327
  **Metric**: Word Error Rate (**WER**)
328
 
329
- | **SNR (dB)** | 100 | 50 | 25 | 15 | 10 | 5 | 0 | -5 |
330
- | --------------- | ----- | ----- | ----- | ----- | ----- | ----- | ----- | ------ |
331
- | **`Canary-1b-v2`** | 2.19% | 2.16% | 2.01% | 2.17% | 2.29% | 2.80% | 5.08% | 19.38% |
332
 
333
 
334
  ### Hallucination Robustness
@@ -346,8 +346,8 @@ Number of characters per minute on [MUSAN](https://www.openslr.org/17) \[16] 48
346
 
347
  | **Dataset** | **WER ↓** |
348
  | ----------------------- | --------- |
349
- | Earnings-22 | TBD |
350
- | This American Life | TBD |
351
 
352
  **Note:** Presented WERs do not include Punctuation and Capitalization errors.
353
 
 
42
 
43
  ### Automatic Speech Recognition (ASR)
44
 
45
+ ![ASR WER Comparison](plots/asr.png)
46
 
47
  *Figure 1: ASR WER comparison across different models. This does not include Punctuation and Capitalisation errors.*
48
 
 
52
 
53
  #### X → English
54
 
55
+ ![AST X-En Comparison](plots/x_en.png)
56
 
57
  *Figure 2: AST X → En COMET scores comparison across different models*
58
 
59
  #### English → X
60
 
61
 
62
+ ![AST En-X Comparison](plots/en_x.png)
63
 
64
  *Figure 3: AST En → X COMET scores comparison across different models*
65
 
 
283
 
284
  | **WER ↓** | Fleurs-25 Langs | CoVoST-13 Langs | MLS - 6 Langs |
285
  | --------------- | -------------------- | -------------------- | ------------------ |
286
+ | **`Canary-1b-v2`** | 8.40% | 8.85% | 7.27% |
287
 
288
 
289
  **Note:** Presented WERs do not include Punctuation and Capitalization errors.
 
326
  Performance across different Signal-to-Noise Ratios (SNR) using MUSAN music and noise samples \[16] on the [LibriSpeech Clean test set](https://www.openslr.org/12).
327
  **Metric**: Word Error Rate (**WER**)
328
 
329
+ | **SNR (dB)** | 100 | 10 | 5 | 0 | -5 |
330
+ | --------------- | ----- | ----- | ----- | ----- | ----- |
331
+ | **`Canary-1b-v2`** | 2.18% | 2.29% | 2.80% | 5.08% | 19.38% |
332
 
333
 
334
  ### Hallucination Robustness
 
346
 
347
  | **Dataset** | **WER ↓** |
348
  | ----------------------- | --------- |
349
+ | Earnings-22 | 13.51% |
350
+ | This American Life | 8.65% |
351
 
352
  **Note:** Presented WERs do not include Punctuation and Capitalization errors.
353
 
plots/asr.png ADDED

Git LFS Details

  • SHA256: 9ed105e2323d620f77c90f8084494b2db2f7916dc99975b0f13c0d1523b985c5
  • Pointer size: 131 Bytes
  • Size of remote file: 136 kB
plots/en_x.png ADDED

Git LFS Details

  • SHA256: 49967add7a65e021eec831cba15f45904562f694535d155f20a31c46d4a420da
  • Pointer size: 131 Bytes
  • Size of remote file: 122 kB
plots/x_en.png ADDED

Git LFS Details

  • SHA256: a444a795dca8fc049a4d4702ca94820cf576c732466d7ffcf050335f6b7c9db9
  • Pointer size: 131 Bytes
  • Size of remote file: 121 kB