Repository navigation
refactor: run audio commands on their executors and the shared HTTP layer - #358
Merged
Merged
Conversation
…ayer `vibe audio transcribe/isolate/clone-voice/dub` and `vibe generate music/sound-effect` each re-implemented the executor the MCP tools and pipelines use, and the copies had drifted. - The commands now validate, call their executor, and format; executors read keys from the config file too (they read only env vars, so keys saved by `vibe setup` were invisible to MCP callers) and return error kinds that map to the CLI envelope. - ElevenLabs sound effects, music, isolation, and voice cloning, and every Replicate call, go through `providerRequest` (retries, classified errors); isolation names the upload from the audio's bytes. Fixes: - `audio transcribe --json -o out.srt` reported the path but never wrote the file. - `audio dub --json` printed the analysis table before the JSON. - `generate music -p replicate -m <variant>` ignored the variant in the executor. - `clone-voice` suggested `vibe audio tts`, which does not exist; it now suggests `vibe generate narration -p elevenlabs --voice <id>`. - `clone-voice --labels` with bad JSON crashed instead of a usage error; `sound-effect` validates duration and prompt influence. - Leftover blank-line runs from the transcription refactor are collapsed. Claude-Session: https://claude.ai/code/session_011S1BLTaFctZNQzGERpkCuA
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
This branch was successfully deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Why
vibe audio transcribe/isolate/clone-voice/dubandvibe generate music/sound-effecteach reimplemented the executor that the MCP tools and pipelines use, and the copies had drifted apart.What changes
vibe setupwere invisible to MCP callers. They now also read the config file.providerFailuremaps to the CLI envelope.providerRequest(retries, classified errors). Isolation names the upload from the audio's bytes.Fixes
audio transcribe --json -o out.srtreported the path but never wrote the file.audio dub --jsonprinted the analysis table before the JSON.generate music -p replicate -m <variant>dropped the variant in the executor path.clone-voicesuggestedvibe audio tts, which doesn't exist. It now suggestsvibe generate narration -p elevenlabs --voice <id>.clone-voice --labelswith bad JSON crashed instead of giving a usage error, andsound-effectnow validates duration and prompt influence.Known issue, next PR
audio dubtranslates by calling Claude's storyboard analysis instead of a translation, and concatenates per-segment audio with no timing. This PR only moves it onto its executor. The translation and timing fix is next.Verification
pnpm build,lint,typecheck, tests (providers 244, CLI 1343 including integration, MCP 73),providers:check, pre-push gate.generate sound-effect -d 2: 2.0 s.audio transcribe --json -o out.srt: the SRT is written with the right cue.generate music -d 3: 3.0 s.audio isolateon a 3.8 s file: classified usage error "below the minimum of 4.6 seconds". On a 7.5 s file: 7.5 s of vocals.https://claude.ai/code/session_011S1BLTaFctZNQzGERpkCuA