KoboldCpp adds YuE2 music and more speech models
KoboldCpp released version 1.124 on October 31, adding YuE2 music generation and more speech models. Its MusicUI now brings transcription alongside generation.
Spec sheet
- Version
- 1.124
- Music
- YuE2
- Speech additions
- BreezeTTS 2, Qwen3-TTS, others
The release incorporates selected audio.cpp integrations, including speech models supporting voice cloning or instructions. YuE2 configurations offer supplied lyrics or an accompanying language model that writes them. The new transcription tab uses Whisper models to produce reference transcripts for speech workflows.
Sources
- koboldcpp-1.124 · KoboldCpp · 31 Oct