6 posts
OxiWhisper 0.1.2 adds a full speaker-diarization pipeline (VAD, clustering, RTTM/DER export, optional ECAPA-TDNN embeddings), word-level timestamps, a translation task, and temperature-fallback decoding — the sovereign speech-recognition layer for the COOLJAPAN ecosystem, still zero C or Python.
OxiAudio is the COOLJAPAN Pure-Rust audio codec and DSP layer — decode FLAC/WAV/MP3/Vorbis/AAC/ALAC/Opus/WavPack/Musepack/MIDI and encode WAV/RF64/FLAC/AIFF/AU with no FFI, plus a full DSP toolkit (filters, dynamics, reverb, phase vocoder, pitch detection, EBU R128 loudness, MFCC/STFT). A clean, memory-safe alternative to libFLAC, libvorbis, libopus, and dr_libs — part of the NoFFI / COOLJAPAN sovereign Rust stack.
OxiSound is the COOLJAPAN Pure-Rust audio device I/O layer — cross-platform playback and capture over ALSA, CoreAudio, and WASAPI, plus MIDI, SMF, and OSC. A NoFFI replacement for cpal/PortAudio and raw ALSA/CoreAudio/WASAPI bindings, part of the sovereign Rust stack.
OxiWhisper 0.1.1 adds transparent GGUF loading, an optional rayon-parallel attention feature, memory-mapped models, f16 KV-cache, FLAC/OGG/MP3/AAC/Opus decoding, and true banded-DTW word timestamps — still Pure Rust, no C or Python.
OxiFFT 0.1.4, the Pure Rust FFT and rustfft replacement, adds a full signal-processing module — Hilbert transform, Welch PSD, cepstrum, and FFT resampling — plus mel-spectrogram and MFCC audio analysis. No C, no Fortran, no FFTW.
OxiWhisper 0.1.0 is a Pure Rust OpenAI Whisper inference engine — GGML loading, Q4_0/Q5_0/Q8_0 quantized inference, beam search, 99-language detection, streaming, and SRT/VTT export, with zero C/C++/Python dependencies.