Automatic Speech Recognition
MLX
English
apple-silicon
speech-to-text
asr
stt
low-bit
ternary
quantization-aware-training
on-device
streaming
Instructions to use FermionResearch/Phonon-1-Micro with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use FermionResearch/Phonon-1-Micro with MLX:
# Download the model from the Hub pip install huggingface_hub[hf_xet] huggingface-cli download --local-dir Phonon-1-Micro FermionResearch/Phonon-1-Micro
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
- Atomic Chat
File size: 2,363 Bytes
84303e4 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 | {
"config_schema": "fermion.phonon/1",
"model_id": "FermionResearch/Phonon-1-Micro",
"name": "Phonon-1 Micro",
"summary": "The smallest Phonon model, for the smallest install. It trades some accuracy for size.",
"license": "apache-2.0",
"library_name": "mlx",
"pipeline_tag": "automatic-speech-recognition",
"language": [
"en"
],
"base_model": "Qwen/Qwen3-ASR-0.6B",
"base_model_relation": "quantized",
"audio": {
"sample_rate": 16000,
"channels": 1,
"designed_utterance_seconds": [
0.5,
30.0
],
"punctuation": "native"
},
"runtime": {
"platform": "macOS on Apple silicon (arm64)",
"framework": "mlx"
},
"artifact": {
"profile": "micro",
"backend": "micro",
"filename": "phonon-micro.bps.tar.zst",
"download_bytes": 285083584,
"sha256": "8a03e55d78e65c10f6ce22aff7e32359ca6a7aaa27973435b9ebc3cdbff13f9c",
"unpacked_bytes": 331234013
},
"profiles": {
"default": "micro",
"listed": [
"micro"
],
"entries": [
{
"key": "micro",
"name": "Micro",
"backend": "micro",
"filename": "phonon-micro.bps.tar.zst",
"download_bytes": 285083584,
"sha256": "8a03e55d78e65c10f6ce22aff7e32359ca6a7aaa27973435b9ebc3cdbff13f9c",
"unpacked_bytes": 331234013,
"listed": true
}
]
},
"family": [
{
"repo_id": "FermionResearch/Phonon-1-Big",
"name": "Phonon-1 Big",
"summary": "The largest Phonon build, and the quickest response of the three. Not the default: it costs 166 MB more than Phonon-1 for no measured accuracy gain."
},
{
"repo_id": "FermionResearch/Phonon-1",
"name": "Phonon-1",
"summary": "The default Phonon model. Statistically tied with Phonon-1 Big on accuracy at 71 % of the download."
},
{
"repo_id": "FermionResearch/Phonon-1-Micro",
"name": "Phonon-1 Micro",
"summary": "The smallest Phonon model, for the smallest install. It trades some accuracy for size."
}
],
"notes": [
"Accuracy and latency figures are published in README.md, not here, so that a card revision can never leave a stale number in machine-readable metadata.",
"The byte size and SHA-256 above describe this release and do not change for it. Check the archive against them before unpacking."
]
}
|