Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
@@ -27,10 +27,12 @@ UTMOS is a MOS prediction system. **A higher UTMOS indicates better quality** of
|
|
27 |
### WER (Word Error Rate)
|
28 |
WER is a common metric for evaluating speech recognition systems. It measures the percentage of words in the generated transcript that differ from the reference (correct) transcript. **A lower WER value indicates higher accuracy**.
|
29 |
Example:
|
|
|
30 |
| Reference | the | cat | sat | on | the | mat |
|
31 |
|-------------|------|-----|---------|-----|------|-----|
|
32 |
| Prediction | the | cat | **sit** | on | the | |
|
33 |
| Label | β
| β
| S | β
| β
| D |
|
|
|
34 |
The WER calculation is done as follows:
|
35 |
```
|
36 |
WER = (S + I + D) / N = (1 + 0 + 1) / 6 = 0.333
|
|
|
27 |
### WER (Word Error Rate)
|
28 |
WER is a common metric for evaluating speech recognition systems. It measures the percentage of words in the generated transcript that differ from the reference (correct) transcript. **A lower WER value indicates higher accuracy**.
|
29 |
Example:
|
30 |
+
|
31 |
| Reference | the | cat | sat | on | the | mat |
|
32 |
|-------------|------|-----|---------|-----|------|-----|
|
33 |
| Prediction | the | cat | **sit** | on | the | |
|
34 |
| Label | β
| β
| S | β
| β
| D |
|
35 |
+
|
36 |
The WER calculation is done as follows:
|
37 |
```
|
38 |
WER = (S + I + D) / N = (1 + 0 + 1) / 6 = 0.333
|