v1.0.26

Highlights

• Add speech-to-text with mmx speech transcribe and sdk.speech.transcribe, powered by asr-1.0. Supports language hints, JSON and detailed transcripts, sentence/word timestamps, SRT/WebVTT subtitles, and streamed text. (#262)
• Improve agent detection for the Codex desktop app and embedded runtimes. (#252, #253)
• Show seconds in the quota reset countdown. (#255)
• Correctly classify search input and output content-filter errors. (#256)
• Include the README in the npm package. (#250)

Try speech-to-text

mmx speech transcribe --file meeting.mp3
mmx speech transcribe --file meeting.mp3 --language zh
mmx speech transcribe --file meeting.mp3 --response-format srt --out meeting.srt

For detailed timestamps and speaker information, use --response-format verbose_json --output json. When piping streamed text, use --stream --output text.

Upgrade

npm install -g mmx-cli@1.0.26
mmx --version

What's Changed

• fix: include README in npm package by @SaladDay in #250
• fix: detect Codex runtime in desktop app by @SaladDay in #252
• fix: detect embedded agent runtimes by @SaladDay in #253
• fix(quota): show seconds in reset countdown by @SaladDay in #255
• fix(search): map output sensitivity to content filter exit code by @renaissance04 in #256
• feat: add mmx speech transcribe for speech-to-text (asr-1.0) by @Wzdhehe in #262

New Contributors

• @renaissance04 made their first contribution in #256
• @Wzdhehe made their first contribution in #262

Full Changelog: v1.0.25...v1.0.26