Poor transcription quality
Symptoms
Section titled “Symptoms”The transcript contains incorrect words, names, numbers, language, or large sections with little useful speech.
Likely causes
Section titled “Likely causes”Recording quality, overlapping speakers, background noise, language selection, source-media quality, and model choice can all affect the output. The application does not provide verified speaker diarization.
Fast checks
Section titled “Fast checks”Compare the transcript with the original recording, confirm the selected language, and identify the current Whisper model. Check whether the source has clear speech rather than only silence or music.
Resolution
Section titled “Resolution”Retry with a deliberate language or another installed Whisper model using a safe representative sample. Correct important content manually and use the recording or another authoritative source for names, decisions, and numbers.
Data-loss warning
Section titled “Data-loss warning”Keep the original recording. Do not overwrite your only accurate source with a generated transcript or reset the session before preserving what you need.
Information to collect for support
Section titled “Information to collect for support”Collect language selection, model identifier, safe audio characteristics, duration, error-free completion status, and examples of the quality issue without disclosing sensitive content.