Release Notes

Fixes

• Fixed an issue where WAV files using a compression format other than standard 16-bit PCM (for example G.722) failed to process. These files are now converted automatically and transcribed as normal.

• If a job’s audio can’t be prepared for transcription, the job now fails straight away and reports the problem clearly, rather than failing later with a less helpful error.

• Fixed an issue where a job that failed while processing its audio could cause the next job to stall. Failed jobs are now cleaned up fully, so later jobs are unaffected.

Improvements

• A new Melia 1 model improves transcription accuracy across many languages including a 1.6% word error rate improvement in English and stronger recognition of medical terminology.

Fixes

• The output_locale property now accepts either a single locale or a list of locales — for example ["en-GB", "cmn-Hans"] — with at most one per language.

New

Oak 1, a new medical model, is now available for batch transcription. It delivers improved medical keyword recognition compared to Melia 1, with the same base accuracy.

Improvements

• A new Melia 1 model improves transcription accuracy across many languages, including a 1.6% word error rate improvement in English and stronger recognition of medical terminology.

• Faster transcription for Melia 1.