Commit
Β·
d889b2b
1
Parent(s):
120634c
Update README.md
Browse files
README.md
CHANGED
|
@@ -52,4 +52,13 @@ transcription = processor.batch_decode(predicted_ids, skip_special_tokens=False)
|
|
| 52 |
transcription = processor.batch_decode(predicted_ids, skip_special_tokens=True)
|
| 53 |
# ['ζ¨ζγγγ«ι»θ©±γθ²Έγγ¦γγγγΎγγγ']
|
| 54 |
|
| 55 |
-
```
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 52 |
transcription = processor.batch_decode(predicted_ids, skip_special_tokens=True)
|
| 53 |
# ['ζ¨ζγγγ«ι»θ©±γθ²Έγγ¦γγγγΎγγγ']
|
| 54 |
|
| 55 |
+
```
|
| 56 |
+
|
| 57 |
+
## Test Result
|
| 58 |
+
|
| 59 |
+
In the table below I report the Character Error Rate (CER) of the model tested on [TEDxJP-10K](https://github.com/laboroai/TEDxJP-10K) dataset.
|
| 60 |
+
| Model | CER |
|
| 61 |
+
| ------------- | ------------- |
|
| 62 |
+
| Ivydata/whisper-small-japanese | **23.10%** |
|
| 63 |
+
| Ivydata/wav2vec2-large-xlsr-53-japanese | **27.87%** |
|
| 64 |
+
| jonatasgrosman/wav2vec2-large-xlsr-53-japanese | 34.18% |
|