Skip to content

v4.0.3 — generate_sfx via Suno fallback, TTS voiceId fix

Choose a tag to compare

View all tags
@elibarnett elibarnett released this 11 Jun 19:41
· 32 commits to main since this release
v4.0.3
8691cf9
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Patch release: both audio-generation paths that kie.ai broke upstream are working again. Both fixes verified with live end-to-end calls over MCP stdio on 2026-06-11.

Fixed

  • generate_sfx was completely broken — kie.ai silently removed elevenlabs/sound-effect-v2 (createTask accepted the slug but every generation failed server-side with code 500, 0 credits consumed). Now routes through Suno V5's sound generator with the same tool name and input schema. Verified: 2 valid 48 kHz MP3s in ~27s. (#10, closes #11)
  • generate_sounds / generate_sfx "done but no results" on success — the sounds operation nests results under data.response.sunoData; pollSunoTask now lifts it to the top, fixing all 14 Suno call sites. (#10, closes #12)
  • generate_tts failed without a voice — kie.ai now returns 422: voiceId cannot be empty when no voice is sent; the tool now always sends a voice ID, and ElevenLabs pricing corrected to per-1000-char rates. Verified: no-voice call produced a valid MP3 in ~21s. (#9, closes #8)

Full details in CHANGELOG.md.

Assets 2
Loading