v0.9.0

speech, speech.cpp's one executable (tts, asr, vad, voice, info, devices, models, pull, rm, quantize, update, serve and worker), and the shared library libspeech with its C API, speech.h, in one archive for each of macOS arm64 (Metal), Windows x64 (Vulkan) and Linux x64 (Vulkan, and the CPU alone; glibc 2.34 or later). The README lists the model families and where their converted weights are on Hugging Face.

What's Changed

• Refuse a GGUF file of another program as written for it, not for an earlier release of speech.cpp by @nyosegawa in #184
• Compare a split Qwen3-ASR dump's stop and merged languages in speech-api-check transcribe by @nyosegawa in #187
• Diarize speakers with Nemotron 3 Diarization through speech_diarize(), diarizations of audio as it arrives and speech diarize by @nyosegawa in #186
• Give speakers in speech serve's Realtime completions and diarized_json, and in speech asr --diarize by @nyosegawa in #188
• Release 0.9.0, with speaker diarization by Nemotron 3 Diarization and speakers on speech serve by @nyosegawa in #189

Full Changelog: v0.8.11...v0.9.0