Two transcription engines. One choice.

convert.express lets you pick the AI model behind your transcription. Choose Whisper for maximum language coverage, or MAI-Transcribe 1.5 for best-in-class accuracy and speed. Same price either way.

Side-by-side comparison

	Whisper	MAI-Transcribe 1.5Preview
Word Error Rate	~5–8%	★2.4% (#3 globally)
Speed (1 hr audio)	~60–90 s	★< 15 s
Languages supported	★57	43
Translation	★English only	Not supported
Entity biasing	Vocabulary hints (free text)	★Up to 200 phrases
Speaker labeling	Yes	Yes
Status	★Generally available	Public preview ★
Price	€0.02 / min	€0.02 / min

Whisper

Word Error Rate~5–8%
Speed (1 hr audio)~60–90 s
Languages supported57
TranslationEnglish only
Entity biasingVocabulary hints (free text)
Speaker labelingYes
StatusGenerally available
Price€0.02 / min

MAI-Transcribe 1.5Preview

Word Error Rate2.4% (#3 globally)
Speed (1 hr audio)< 15 s
Languages supported43
TranslationNot supported
Entity biasingUp to 200 phrases
Speaker labelingYes
StatusPublic preview ★
Price€0.02 / min

When to use each model

Whisper

Rare or regional languages not covered by MAI
Mixed-language audio including languages outside MAI's 43
Production workflows where you need a GA-stable model
Languages like Hebrew, Persian, Swahili, Welsh, and more

MAI-Transcribe 1.5

Preview

Maximum accuracy — 2.4% WER, #3 globally
Speed-critical workflows — 1 hr of audio in under 15 s
Professional use cases: medical, legal, research (entity biasing)
Word-level timestamps for precise subtitle generation

About the MAI-Transcribe Preview status

MAI-Transcribe 1.5 is currently in Microsoft's public preview programme — it has no formal SLA. If the model is unavailable, convert.express automatically falls back to Whisper so your job always completes. We will promote it to default once it reaches general availability.

Model selection is in Advanced Options on the upload form — free to try.

Try it now →