This release fixes issues for SREs watching stability and regressions.
✓ No known CVEs patched in this version
Topics
+1 more
Summary
AI summaryFixed transcription tools passing an invalid language argument which misread as speaker detection.
Changes in this release
| Type | Severity | Summary | CVE |
|---|---|---|---|
| Bugfix | Medium |
`transcribe_audio` and `transcribe_folder_audio` no longer pass an invalid `language` argument. `transcribe_audio` and `transcribe_folder_audio` no longer pass an invalid `language` argument. Source: llm_adapter@2026-06-05 Confidence: high |
— |
Full changelog
Bug-fix release.
Audio transcription no longer passes an invalid language argument. The transcribe_audio (clip) and transcribe_folder_audio tools in the full (--full) server were calling TranscribeAudio(language) with a language string, but the Resolve scripting API has never accepted a language positional — its signature is TranscribeAudio(useSpeakerDetection=None). The string was silently coerced to a truthy boolean and misread as a speaker-detection flag, and the success message falsely claimed a transcription language. The language is controlled by the project's Speech Recognition setting, not per call.
transcribe_audio(clip_name, use_speaker_detection=None)— thelanguageparameter is replaced with an optionaluse_speaker_detectionboolean (Resolve 21+). Omit it to use the project's setting.transcribe_folder_audio(folder_name, use_speaker_detection=None)— same change for the folder-level tool.- Both pass the boolean through only when supplied, so older Resolve builds (which take no argument) keep working.
The consolidated 32-tool server was unaffected — its folder/media_pool_item transcribe_audio actions already called TranscribeAudio() correctly.
Weekly OSS security release digest.
The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.
No spam, unsubscribe anytime.
Share this release
Related context
Beta — feedback welcome: [email protected]