domains / families / research / interspeech-experiments

interspeech-experiments

separate

Use when designing or auditing the experimental evidence for an INTERSPEECH paper — task-correct metrics (WER/CER, MOS/CMOS, EER/minDCF, PESQ/STOI), baselines reviewers accept, significance testing over utterances and seeds, condition coverage across speakers/noise/languages, and data hygiene for speech corpora.

When to use

• designing experimental evidence for INTERSPEECH papers
• auditing experimental evidence for INTERSPEECH papers
• selecting task-correct metrics
• defining acceptable baselines
• conducting significance testing
• ensuring condition coverage
• checking data hygiene for speech corpora

Skill page

Source