Skip to content

Commit 63d1b38

Browse files
Update articles/ai-services/speech-service/how-to-pronunciation-assessment.md
Co-authored-by: Eric Urban <[email protected]>
1 parent f7d1eb8 commit 63d1b38

File tree

1 file changed

+1
-1
lines changed

1 file changed

+1
-1
lines changed

articles/ai-services/speech-service/how-to-pronunciation-assessment.md

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1034,7 +1034,7 @@ pronunciationAssessmentConfig?.nbestPhonemeCount = 5
10341034
- Pronunciation assessment uses a fixed version of the speech to text model, different from the standard speech to text model.
10351035
- Pronunciation scores are calculated by weighting accuracy, prosody, fluency, and completeness scores based on specific formulas for reading and speaking scenarios.
10361036

1037-
When sorting the scores of accuracy/prosody/fluency/completeness from low to high (if each score is available) and representing the lowest score to the highest score as s0 to s3, the pronunciation score is calculated as follows:
1037+
When sorting the scores of accuracy, prosody, fluency, and completeness from low to high (if each score is available) and representing the lowest score to the highest score as s0 to s3, the pronunciation score is calculated as follows:
10381038

10391039
For reading scenario:
10401040
- With prosody score: PronScore = 0.4 * s0 + 0.2 * s1 + 0.2 * s2 + 0.2 * s3

0 commit comments

Comments
 (0)