maxffarrell/ARK-ASR-0.6B-GGUF

ARK-ASR-0.6B GGUF candidate

Original authors and conversion credits

ARK-ASR is the work of AutoArk and research authors Yu Lin, Yiming Wang, Runyuan Cai, and Xiaodong Zeng. Original model: Edge0/ARK-ASR-0.6B · Original project · Research paper.

This conversion reuses harshav's ARK GGUF converter and runtime from ARK-ASR-3B-GGUF, built on the transcribe.cpp authors' native engine. maxffarrell contributed the 0.6B adaptation, quantization, validation and packaging, with Codex assistance—not the original model or research.

Licenses: original weights are Apache-2.0; converter/runtime are MIT. See full attribution and research citation, NOTICE, weight license and converter license. The v2 GGUF embeds these credits and source links so they travel with the file.

Community Q8 conversion of Edge0/ARK-ASR-0.6B, prepared for a draft Hex integration. No retraining. Not affiliated with the original authors, and not a canonical handy-computer artifact.

Requires an ARK-enabled runtime. Stock llama.cpp, whisper.cpp and released transcribe-cpp 0.2.4 cannot load it. Use the candidate runtime at maxffarrell/transcribe.cpp, revision 19f301609c06126e5071d1de5acdd95c714eec05. Upstream draft PR.

File

ark-asr-0.6b-Q8_0-v2.gguf: 1,372,882,944 bytes. SHA-256: 20a9954ce65e0e724a836194a12ad10879ca746c133dcea54e0d15f2132a94bd.

The 0.6B name describes the decoder; the full model also contains the audio encoder and adapter. Q8 linear weights, F16 tied embedding/output matrix and convolutions, F32 norms/biases/frontend tensors.

Run

git clone https://github.com/maxffarrell/transcribe.cpp
cd transcribe.cpp
git checkout 19f301609c06126e5071d1de5acdd95c714eec05
cmake -S . -B build -DCMAKE_BUILD_TYPE=Release -DTRANSCRIBE_BUILD_EXAMPLES=ON
cmake --build build --target transcribe-cli -j 8
hf download maxffarrell/ARK-ASR-0.6B-GGUF ark-asr-0.6b-Q8_0-v2.gguf --local-dir models/ark
build/bin/transcribe-cli --backend metal -m models/ark/ark-asr-0.6b-Q8_0-v2.gguf audio.wav

Use --backend cpu for CPU inference. Input frontend: mono 16 kHz. Long recordings use overlapping windows of at most 30 seconds. Language is inferred from audio; forced-language steering, translation, streaming and timestamps are not supported by this candidate.

Validation and limitations

Six public short samples from transcribe.cpp (JFK, Mandarin, Japanese, Korean, noise and silence) produced exactly matching text in the original PyTorch checkpoint, F16 GGUF and this Q8 GGUF. The Q8 JFK transcript also matched on CPU and Metal. A 197-second public English speech clip completed through the runtime's overlapping-window path on Metal. The existing 45 runtime CTests passed. See validation.json for the compared text.

Reference agreement is not a ground-truth WER measurement. The original model and both GGUF precisions emitted Japanese text for the Korean sample and 嗯。 for noise and silence. Automatic language detection and hallucination on nonspeech remain material limitations. The publisher's 19-language list is coverage metadata, not validation of all languages. Long-audio completion does not establish parity against an unchunked reference. Full tensor parity, WER evaluation and upstream port acceptance remain incomplete. Treat this as an experimental model, not an accuracy or speed recommendation.

Reproduce

Source: Edge0/ARK-ASR-0.6B@45776b56d58cdfb2e2eb632f7e110f38684633e0. Original model.safetensors SHA-256: 57a86ce1c2f2c2d6ebb7ad9642c9e951a5109625122b48c9180126a28787673d.

uv run convert-arkasr-to-gguf.py --input Edge0/ARK-ASR-0.6B \
  --revision 45776b56d58cdfb2e2eb632f7e110f38684633e0 \
  --variant ark-asr-0.6b --outtype q8_0 --output ark-asr-0.6b-Q8_0-v2.gguf

The included converter reuses harshav's work from ARK-ASR-3B-GGUF, with variant, revision and capability metadata adapted for 0.6B. Its source archive is pinned in conversion.json. Converter/runtime license: MIT (CONVERTER-LICENSE); weights: Apache-2.0 (LICENSE). Original architecture, training and evaluation: AutoArk/open-audio-opd.

This community conversion and its draft integration were prepared with Codex at the contributor's request.

Earlier artifact

ark-asr-0.6b-Q8_0.gguf is retained unchanged for existing pinned consumers. Use ark-asr-0.6b-Q8_0-v2.gguf for embedded attribution. Every tensor in v2 is byte-identical to the earlier file; only descriptive metadata changed.