Instructions to use basyauma-tech/Haas-ja-v1.0-mlx with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use basyauma-tech/Haas-ja-v1.0-mlx with MLX:
# Download the model from the Hub pip install huggingface_hub[hf_xet] hf download basyauma-tech/Haas-ja-v1.0-mlx --local-dir Haas-ja-v1.0-mlx
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
- Atomic Chat
Haas-ja-v1.0-mlx
้ๅ ฌๅผใฎๅคๆ็ฉใงใใ third-intelligence/Haas-ja-v1.0ใฎMLX็ใงใใThird Intelligence, Inc.ใจใฏๆๆบใปๆฟ่ชใปๅพๆดใฎ้ขไฟใซใใใพใใใ
้ใฟใฏๅ
ใฎbf16ใฎๅคใไฟๆใใ่ปธใฎไธฆในๆฟใใจๆจ่ซใซไธ่ฆใชnum_batches_trackedใฎ้คๅปใ ใใ่กใใพใใHugging Faceไธใฎๅ้กใฏquantizedใงใใใ้ๅญๅใใฆใใพใใใ
ๆฆ่ฆใป็ฐๅข
ๅ ใขใใซใฏFastConformerใจใณใณใผใใจTDTใใณใผใ๏ผไบๆธฌใใใใฏใผใฏใฏKimi Delta Attention๏ผใงใๅ ฌๅผใฎๅฎ่ก็ฐๅขใฏLinuxใจNVIDIA GPU๏ผNeMo 2.7๏ผใซ้ใใใพใใ
- ใใฎๅคๆ็ฉ: Apple SiliconใฎMacๅฐ็จใNeMoใปCUDAใปTritonใฏไธ่ฆใๆจ่ซใณใผใใฏๅฅ้haas-ja-localใไฝฟ็จใ
- Python 3.12๏ผ>=3.12,<3.13๏ผใๅฎ่กๆไพๅญใฏmlxใปnumpyใปsentencepieceใปsoundfileใปsoxrใtorchไธ่ฆใ้ใฟ็จใซ็ด1.2GBๅฟ ่ฆใ
- ๆค่จผๆฉใฏApple M5 Max๏ผ128GB๏ผใmacOS 27.2ใmlx 0.32.4ใฎ1ๅฐใฎใฟใ
ไฝฟใๆน
hfใฏhuggingface_hubใซไปๅฑ๏ผๅฟ
่ฆใชใpip install huggingface_hub๏ผใ
pip install "haas-ja-local[mlx] @ git+https://github.com/basyauma-tech/haas-ja-local"
hf download basyauma-tech/Haas-ja-v1.0-mlx --local-dir Haas-ja-v1.0-mlx
Python:
from haas_ja_local import load_audio, load_model, transcribe
backend, tokenizer = load_model("Haas-ja-v1.0-mlx")
result = transcribe(backend, tokenizer, load_audio("audio.wav"))
print(result.text)
CLI:
haas-ja-local transcribe Haas-ja-v1.0-mlx audio.wav
- ๅ ฅๅ: soundfileๅฏพๅฟๅฝขๅผ๏ผWAVใปFLACใชใฉ๏ผใ่คๆฐใใฃใณใใซใฏๅนณๅใใฆใขใใฉใซๅใใsoxrใง16kHzใซๅคๆใ
- ้ทๆ้้ณๅฃฐ: 30็ง่ถ ใ้ๆฌกๅๅฒใๅๆญ็ใฎๅ ้ ญใใ30็งยฑ2็งๅ ใงใๆใ้ใใช0.1็ง็ชใฎไธญๅฟใๅๆญ็นใจใใ่ช่ญ็ตๆใ้ฃ็ตใ
- Metalใญใฃใใทใฅ: ๆขๅฎ1GiBใ
HaasMLX.loadใฎcache_limit_bytesใงๅคๆด๏ผNoneใฏ่จญๅฎใใชใ๏ผใ - ๆๅป:
--json --timestampsใง1ใใกใคใซ1่กใฎJSONใใใผใฏใณใใจใฎ้ๅงใป็ตไบ็งใ80msๅปใฟใงๅบๅใ่คๆฐใใคใใใผใฏใณใฎๆๅญใฏ1ไปถใซ็ตฑๅใ
haas-ja-local transcribe Haas-ja-v1.0-mlx audio.wav --json --timestamps
{"file": "audio.wav", "text": "ไปๆฅใฏ่ฏใๅคฉๆฐใงใใญใ", "tokens": [{"text": "ไป", "start": 0.0, "end": 0.32}, {"text": "ๆฅใฏ", "start": 0.32, "end": 0.64}, {"text": "่ฏ", "start": 0.64, "end": 0.88}, {"text": "ใ", "start": 0.88, "end": 1.04}, {"text": "ๅคฉ", "start": 1.04, "end": 1.28}, {"text": "ๆฐ", "start": 1.28, "end": 1.6}, {"text": "ใงใใญ", "start": 1.6, "end": 1.84}, {"text": "ใ", "start": 1.84, "end": 2.0}]}
็ฒพๅบฆ
asr-ja-bench v1.1ใjacer v1.2.0ใงๆธฌๅฎใๅไฝใฏ%ใ
| ใตใใปใใ | ็บ่ฉฑๆฐ | ้ณๅฃฐ๏ผๆ้๏ผ | ๆๆจ | ใใฎๅคๆ | ๅ ฌ่กจๅค | ๅทฎ๏ผใใคใณใ๏ผ |
|---|---|---|---|---|---|---|
| FLEURS | 650 | 2.36 | CER strict | 3.65 | 3.65ใ3.67 | 0.00 |
| FLEURS | 650 | 2.36 | CER flex | 2.98 | ๆช็ งๅ | - |
| FLEURS | 650 | 2.36 | CER kana | 2.33 | ๆช็ งๅ | - |
| JVNV | 1,375 | 3.33 | CER strict | 4.51 | 4.50ใ4.53 | 0.00 |
| JVNV | 1,375 | 3.33 | CER flex | 3.06 | ๆช็ งๅ | - |
| JVNV | 1,375 | 3.33 | CER kana | 2.25 | ๆช็ งๅ | - |
ๆธฌๅฎๆกไปถ:
rebuild.pyใง่ฉไพกใใผใฟใๅๆง็ฏใใๅ จๆธใ่ตทใใใฎใใใทใฅ็ งๅใซๅๆ ผใJVNVใฎ16kHzๅใฏๆณๅฎใฎffmpeg 7.1ใงใฏใชใ9.0.2ใฎใใใๅ ฌๅผ้ณๅฃฐใจๅพฎๅทฎใฎๅฏ่ฝๆงใใใ- 1็บ่ฉฑใใค่ช่ญๅพใใตใใปใใๅ
จไฝใง
jacer.cerใ1ๅ่จ็ฎ๏ผใใคใฏใญๅนณๅ๏ผใ3ๆๆจใจใๅฅ่ชญ็นใป็ฉบ็ฝใ้คๅคใๅ จ็บ่ฉฑ30็งไปฅไธใงๅๅฒใชใใ - ไธ่จM5 MaxใMetalใใใใ1ใใจใณใณใผใbf16ใไบๆธฌใป็ตๅใใใใฏใผใฏfloat32ใ
- ๅ ฌ่กจๅคใฏๅ ฌๅผasr-ja-bench v1.1๏ผNeMoใปbf16ใปH200๏ผใใขใใซใซใผใใจๅ ฌ้่จไบใฎๅฐๆฐ็ฌฌ2ไฝใฎๅทฎใๅน ใง่กจ็คบใๅทฎใฏ่ฟใ็ซฏใจใฎๅทฎใ็ฏๅฒๅ ใฏ0ใflexใปkanaใฏๆช็ งๅใ
- ็ฒพๅบฆ่ชฟๆดใชใใฎๅๅๅฎ่กใ
ๆชๆธฌๅฎ: JSUT basic5000ใปTEDxJP-10KใปSpeechBSD๏ผ5ใตใใปใใใฎ้ใฟไปใCERใชใ๏ผใIntel Macใใกใขใช16GBไปฅไธใใใใ2ไปฅไธใๅไธ่ฉฑ่ ใ่ฉฑใ็ถใใๅฎ้ใฎ้ทๆ้้ฒ้ณใ
่ฉณ็ดฐใปๅ็พๆ้ : ็ฒพๅบฆใจ้ๅบฆใๆฐๅคไธ่ดใ
MLX็๏ผbf16๏ผใจPyTorchๅ็ งๅฎ่ฃ ๏ผfloat32๏ผใฎๆฏ่ผใงใฏใFLEURS650็บ่ฉฑไธญ632็บ่ฉฑ๏ผ97.2%๏ผใงใใผใฏใณๅใจๆฌๆใไธ่ดใใ็ฐใชใ18็บ่ฉฑใฎใใก12็บ่ฉฑใฏๅฅ่ชญ็นใ ใใฎ้ใใงใใใ
ๅฎ่ฃ ใฎ็ขบ่ช
PyTorchๅ็ งๅฎ่ฃ ใฏใๅใใใงใใฏใใคใณใใใ็ฌฌไธ่ ใๆธใๅบใใfloat32ใฎONNXใจFLEURSใฎ6็บ่ฉฑใงๆฏ่ผใใๆๅคง่ชคๅทฎใฏlog-mel 9.54e-07ใใจใณใณใผใๅบๅ๏ผ็ธๅฏพ๏ผ4.36e-06ใ็ตๅใใใใฏใผใฏๅบๅ๏ผ็ธๅฏพ๏ผ1.16e-06ใใใผใฏใณๅใฏ6็บ่ฉฑใในใฆไธ่ดใงใใใ
้ๅบฆใจใกใขใช
FLEURSๅ จ650็บ่ฉฑใใๅ ้ ญ3็บ่ฉฑใฎใฆใฉใผใ ใขใใใ้คๅคใใ647็บ่ฉฑใ2.35ๆ้ใไธ่จ็ฐๅขใป่จ็ฎๅใใใใ1ใง้ๆฌก่ช่ญใ้ณๅฃฐ่ชญใฟ่พผใฟใ้คใ่ช่ญๆ้ใGB๏ผ10^9ใใคใใ
| ้ ็ฎ | ๆขๅฎ๏ผใญใฃใใทใฅไธ้1GiB๏ผ | ไธ้ใชใ๏ผๅ่ใๆณจ๏ผ |
|---|---|---|
| RTFx๏ผ1็บ่ฉฑใใจใฎไธญๅคฎๅค๏ผ | 254.2 | 250.9 |
| RTFx๏ผๅ่จ้ณๅฃฐรทๅ่จๅฆ็ๆ้๏ผ | 256.1 | 255.9 |
| ใใผใฏRSS | 1.57GB | 2.68GB |
mlxใฎใใผใฏๅฒใๅฝใฆ๏ผmx.get_peak_memory()๏ผ |
1.76GB | 1.82GB |
| macOSใฎใใผใฏใกใขใชใใใใใชใณใ | 3.19GB | 14.40GB |
- ไธ้ใชใใฎใใใใใชใณใใฎใฟๅๆกไปถใฎๅๅฎ่กๅคใไปใฏๅๅใ
- ไธ้ใชใ: 1็บ่ฉฑใฎไธญๅคฎๅค0.050็งใใขใใซ่ชญใฟ่พผใฟ0.15็ง๏ผOSใญใฃใใทใฅๆธใฟ๏ผใ
- CLIใง33.8ๅ๏ผ2,026.5็ง๏ผใ68ๅๅฒ: ่ตทๅใปใขใใซ่ชญใฟ่พผใฟ่พผใฟ6.97็งใๆๅคงRSS 1.74GBใใใผใฏใใใใใชใณใ3.48GBใ
- ๅ ใขใใซใฎๅ ฌ่กจRTFxใฏNVIDIA H200ใปใใใ512ใฎใใๆฏ่ผไธๅฏใ
- ๅใ33.8ๅ้ณๅฃฐใงไธ้ใชใใงใฏใใใใใชใณใ13.96GBใๅฎไฝฟ็จใฏ1.25GBใงไธๅฎใ่งฃๆพๆธใฟMetalใใใใกใฎไฟๆใงใใใชใผใฏใงใฏใชใใๆขๅฎใฎไธ้ใฏใใฎๅขๅ ใๆใใใใใ
ๅคๆๆนๆณ
pip install "haas-ja-local[convert] @ git+https://github.com/basyauma-tech/haas-ja-local"
hf download third-intelligence/Haas-ja-v1.0 Haas-ja-v1.0.nemo --revision 660728d39019e5e065046154beca44ac9a66a540 --local-dir upstream
haas-ja-local convert-mlx --nemo upstream/Haas-ja-v1.0.nemo --out Haas-ja-v1.0-mlx
| ้ ็ฎ | ๅค |
|---|---|
| ๅคๆๅ | third-intelligence/Haas-ja-v1.0ใฎHaas-ja-v1.0.nemo |
| ใชใใธใงใณ | 660728d39019e5e065046154beca44ac9a66a540๏ผ2026-11-08ๅๅพ๏ผ |
| SHA-256 | c6ed38698300a371a86eadab4b56d5458f47623f7fc4a5f1e938b5adbe8458aa |
| ใตใคใบ | 1,232,384,000ใใคใ |
| ๅคๆใใผใซ | haas-ja-local 0.1.0๏ผmodel.safetensorsใฎใกใฟใใผใฟใซ่จ้ฒ๏ผ |
- bf16ใฎๅคใปๅใฏไธๅคใ็ณใฟ่พผใฟใฎใฟMLXใฎใใฃใณใใซๆซๅฐพ้ ็ฝฎใธๅคๆด: 2ๆฌกๅ [O,I,H,W]โ[O,H,W,I]ใ1ๆฌกๅ ๏ผๆทฑใๆนๅใๅซใ๏ผ[O,I,K]โ[O,K,I]ใ
- ๅๅคๆใฎใใณใฝใซๅคใฏใใใไธ่ดใsafetensorsใฎใกใฟใใผใฟใญใผ้ ใฏไธๅฎใฎใใใใใกใคใซๅ จไฝใฎSHA-256ใฏไธ่ดใใชใๅ ดๅใใใ
ใใกใคใซไธ่ฆง
| ใใกใคใซ | ๅ ๅฎน |
|---|---|
model.safetensors |
bf16ใ1,231,827,296ใใคใใSHA-256: 96600ed4ca8a4d9d45e6ed6bfd276ab9109da3f1c9e20e2bb0870e6f894fe5e4 |
config.json |
ใขใใซ่จญๅฎ |
tokenizer.model |
SentencePiece BPEใ3,072ใใผใฏใณ |
LICENSE |
ใฉใคใปใณใน |
NOTICE |
ๅธฐๅฑๆ ๅ ฑ |
ๅถ้ไบ้
- ๅ ใขใใซๅๆงใใใฃใฉใผ๏ผใใใผใใใใฎใ๏ผใๆใใใใใๆผขๅญใปใใชใปๆฐๅญใฎ่กจ่จๆๅฎใฏไธๅฏใ
- ็ก้ณใป้้ณใง็ญ่ชใๅบใๅฏ่ฝๆงใใใ10็งใฎ็ก้ณใปๅผฑใใใฏใคใใใคใบใฎๅๆ้ณๅฃฐ4็จฎใงใฏๅบๅใชใใๅฎ้ฒ้ณใฏๆช็ขบ่ชใ
- 30็ง่ถ
ใฎๅๅฒใงCERใๆชๅใFLEURS 150็บ่ฉฑใ้ฃ็ตใใ33.8ๅ้ณๅฃฐใฎCER strictใฏใ็บ่ฉฑๅฅ4.10%ใ็ด30็งๅๅฒ6.44%ใPyTorch๏ผfloat32๏ผ6.39%ใMLXใฎๆฐๅค่ชคๅทฎใงใฏใชใใใขใใซใจๅ
ฌๅผ
transcribe_longformใซๆฒฟใฃใๅๅฒ่ฆๅใซใใใ - ๆชๅใฎ็ด1ใใคใณใใฏ็ฐ่ฉฑ่ ใฎ้ฃ็ต๏ผ็บ่ฉฑใๅใใ30็งไปฅๅ ใซใพใจใใฆใ5.08%๏ผใๅไธ่ฉฑ่ ใฎๅฎ้ฒ้ณใงใฏๅฐใใๅฏ่ฝๆงใใใๆฎใ็ด1.4ใใคใณใใฏ็บ่ฉฑ้ไธญใฎๅๆญ๏ผ67ๅข็ไธญ50๏ผใ
- ใใใ1ใฎใฟใในใใชใผใใณใฐใป่ฉฑ่ ใฉใใซใปๅ ฌๅผHaas-ASRใฎใซในใฟใ ่ชๅฝใฏๆชๅฏพๅฟใ
- ๆๅปใฏ80msๅปใฟใฎใใผใฏใณๅไฝใๅ่ชๅไฝใงใฏใชใใ
ใฉใคใปใณในใจๅธฐๅฑ
- ้ใฟใปๆจ่ซใณใผใ: Apache-2.0ใๅ
้ใฟใฏthird-intelligence/Haas-ja-v1.0๏ผCopyright Third Intelligence, Inc.๏ผใๅธฐๅฑใฏ
NOTICEใใHaasใใฏๅคๆๅ ใ็คบใๅ็งฐใ - ๅ่ๅฎ่ฃ : NVIDIA NeMoใthird-intelligence/Haas-ASR๏ผๅ ฌๅผๆจ่ซใใใฑใผใธ๏ผใmatsuurajun/haas-mlxใSanami/haas-ja-onnxใฏApache-2.0ใflash-linear-attentionใฏMITใ
- Sanamiใฎfloat32 ONNXใฏๆฐๅคๆฏ่ผใฎใฟใงใ้ใฟใฏๅ้
ๅธใใชใใ้
ๅธ้ใฟใฏใในใฆๅ
ใฎ
.nemoใฎ้ๅ ฌๅผใฎๅคๆ็ฉใ - ่ฉไพกใใผใฟ: FLEURSใฏCC BY 4.0ใJVNVใฏCC BY-SA 4.0ใๅๆขฑใปๅ้ ๅธใใชใใ
English
This is an unofficial conversion of third-intelligence/Haas-ja-v1.0 to MLX. It is not affiliated with, endorsed by, or sponsored by Third Intelligence, Inc.
Original bf16 values are preserved; only tensor axes are rearranged and unused num_batches_tracked values removed. Hugging Face classifies it as quantized, but no quantization is applied.
Overview and requirements
The original model uses a FastConformer encoder and a TDT decoder (Kimi Delta Attention prediction network); the official runtime is limited to Linux with NVIDIA GPUs (NeMo 2.7).
- This conversion: Apple Silicon Macs only; no NeMo, CUDA or Triton. Inference code is separate: haas-ja-local.
- Python 3.12 (>=3.12,<3.13); mlx, numpy, sentencepiece, soundfile, soxr. No torch needed. Allow about 1.2 GB for weights.
- Tested on one Apple M5 Max (128 GB), macOS 27.2, mlx 0.32.4.
Usage
hf comes with huggingface_hub (pip install huggingface_hub if needed).
pip install "haas-ja-local[mlx] @ git+https://github.com/basyauma-tech/haas-ja-local"
hf download basyauma-tech/Haas-ja-v1.0-mlx --local-dir Haas-ja-v1.0-mlx
haas-ja-local transcribe Haas-ja-v1.0-mlx audio.wav
haas-ja-local transcribe Haas-ja-v1.0-mlx audio.wav --json --timestamps
from haas_ja_local import load_audio, load_model, transcribe
backend, tokenizer = load_model("Haas-ja-v1.0-mlx")
print(transcribe(backend, tokenizer, load_audio("audio.wav")).text)
- Input: soundfile formats (WAV, FLAC, etc.). Channels are averaged to mono; soxr resamples to 16 kHz.
- Long audio: split sequentially above 30 seconds. Each cut is the centre of the quietest 0.1-second window within 30 ยฑ 2 seconds of the piece's start. Results are joined.
- Metal cache: 1 GiB default; change
cache_limit_bytesinHaasMLX.load(Noneleaves it unchanged). - Times:
--json --timestampsemits one JSON line per file, with token start/end seconds in 80 ms steps. Multiple byte tokens for one character are merged.
Accuracy
asr-ja-bench v1.1, jacer v1.2.0; values in %.
| Subset | Utterances | Audio (hours) | Metric | This conversion | Published | Difference (points) |
|---|---|---|---|---|---|---|
| FLEURS | 650 | 2.36 | CER strict | 3.65 | 3.65โ3.67 | 0.00 |
| FLEURS | 650 | 2.36 | CER flex | 2.98 | not compared | - |
| FLEURS | 650 | 2.36 | CER kana | 2.33 | not compared | - |
| JVNV | 1,375 | 3.33 | CER strict | 4.51 | 4.50โ4.53 | 0.00 |
| JVNV | 1,375 | 3.33 | CER flex | 3.06 | not compared | - |
| JVNV | 1,375 | 3.33 | CER kana | 2.25 | not compared | - |
Conditions:
- Data rebuilt with
rebuild.py; all transcript hashes passed. JVNV resampling to 16 kHz used ffmpeg 9.0.2 instead of the expected 7.1, so audio may differ slightly. - Utterances transcribed individually; one
jacer.cercall per subset (micro-average). All three metrics ignore punctuation/spaces. All utterances โค30 seconds; no splitting. - M5 Max setup above, Metal, batch 1; bf16 encoder, float32 prediction/joint networks.
- Published: official asr-ja-bench v1.1 (NeMo, bf16, H200). Model-card/blog differences in the second decimal place form the range. Differences use the nearer endpoint, or 0 within the range. flex/kana not compared.
- First run; no accuracy tuning.
Not measured: JSUT basic5000, TEDxJP-10K, SpeechBSD (no weighted CER across all five subsets); Intel Macs; โค16 GB machines; batch sizes โฅ2; real long recordings of one continuous speaker.
Details/reproduction (Japanese): accuracy and speed, numerical agreement.
MLX (bf16) and the PyTorch reference (float32) matched in both token sequences and text on 632 of 650 FLEURS utterances (97.2%); 12 of the 18 differing utterances differed only in punctuation.
Implementation verification
The PyTorch reference was compared on 6 FLEURS utterances against third-party float32 ONNX exported from the same checkpoint: maximum errors were 9.54e-07 for log-mel, 4.36e-06 for encoder output (relative), and 1.16e-06 for joint network output (relative); token sequences matched on all 6 utterances.
Speed and memory
647 of 650 FLEURS utterances, 2.35 hours, excluding three warm-ups. Same setup/precision as above, batch 1, sequential. Transcription time excludes audio reading. GB = 10^9 bytes.
| Item | Default (1 GiB cache cap) | No cap (reference, see note) |
|---|---|---|
| RTFx (median per utterance) | 254.2 | 250.9 |
| RTFx (total audio / total processing time) | 256.1 | 255.9 |
| Peak RSS | 1.57 GB | 2.68 GB |
mlx peak allocation (mx.get_peak_memory()) |
1.76 GB | 1.82 GB |
| macOS peak memory footprint | 3.19 GB | 14.40 GB |
- Only the uncapped footprint is from a same-condition rerun; other values are from the first run.
- Uncapped: median 0.050 seconds/utterance; loading 0.15 seconds with files in the OS cache.
- CLI, 33.8 minutes (2,026.5 seconds), 68 pieces: 6.97 seconds including startup/loading; maximum RSS 1.74 GB, peak footprint 3.48 GB.
- Published RTFx uses NVIDIA H200, batch 512; not comparable.
- Uncapped on the same 33.8-minute file: footprint 13.96 GB, active memory constant at 1.25 GB. Retained freed Metal buffers, not a leak; the default cap limits this growth.
Conversion
pip install "haas-ja-local[convert] @ git+https://github.com/basyauma-tech/haas-ja-local"
hf download third-intelligence/Haas-ja-v1.0 Haas-ja-v1.0.nemo --revision 660728d39019e5e065046154beca44ac9a66a540 --local-dir upstream
haas-ja-local convert-mlx --nemo upstream/Haas-ja-v1.0.nemo --out Haas-ja-v1.0-mlx
| Item | Value |
|---|---|
| Source | Haas-ja-v1.0.nemo from third-intelligence/Haas-ja-v1.0 |
| Revision | 660728d39019e5e065046154beca44ac9a66a540 (downloaded 2026-11-08) |
| SHA-256 | c6ed38698300a371a86eadab4b56d5458f47623f7fc4a5f1e938b5adbe8458aa |
| Size | 1,232,384,000 bytes |
| Converter | haas-ja-local 0.1.0 (recorded in the model.safetensors metadata) |
- bf16 values/types unchanged. Only convolution axes move to channels-last: 2-D [O,I,H,W]โ[O,H,W,I]; 1-D, including depthwise, [O,I,K]โ[O,K,I].
- Repeated conversion gives bit-identical tensors. File SHA-256 may differ because safetensors metadata-key order is not fixed.
Files
| File | Contents |
|---|---|
model.safetensors |
bf16; 1,231,827,296 bytes; SHA-256: 96600ed4ca8a4d9d45e6ed6bfd276ab9109da3f1c9e20e2bb0870e6f894fe5e4 |
config.json |
Model configuration |
tokenizer.model |
SentencePiece BPE, 3,072 tokens |
LICENSE |
License |
NOTICE |
Attribution |
Limitations
- As with the original model, fillers ("ใใผ", "ใใฎ") may be omitted; kanji/kana/numeral style cannot be selected.
- Silence/noise may produce short words. Four synthetic 10-second silence/weak-white-noise inputs produced none; real recordings untested.
- Splitting above 30 seconds raises CER. For 150 FLEURS utterances joined into 33.8 minutes, CER strict: 4.10% individually, 6.44% with ~30-second splits, 6.39% in PyTorch (float32). This reflects the model and splitting rule following official
transcribe_longform, rather than MLX numerics. - About 1 point comes from joining different speakers (5.08% even when grouping intact utterances into โค30 seconds); this may be smaller with one continuous speaker. About 1.4 points come from mid-utterance cuts (50 of 67 boundaries).
- Batch 1 only. No streaming, speaker labels or official Haas-ASR custom vocabulary.
- Token timestamps in 80 ms steps, not word timestamps.
License and attribution
- Weights/code: Apache-2.0. Source weights: third-intelligence/Haas-ja-v1.0, Copyright Third Intelligence, Inc. Attribution:
NOTICE. "Haas" identifies the source. - References: NVIDIA NeMo, third-intelligence/Haas-ASR (official inference package), matsuurajun/haas-mlx, Sanami/haas-ja-onnx: Apache-2.0. flash-linear-attention: MIT.
- Sanami's float32 ONNX is used only for numerical comparison; its weights are not redistributed. All distributed weights are unofficial conversions of the original
.nemo. - Evaluation data: FLEURS CC BY 4.0; JVNV CC BY-SA 4.0. Neither included nor redistributed.
- Downloads last month
- -
Quantized
Model tree for basyauma-tech/Haas-ja-v1.0-mlx
Base model
third-intelligence/Haas-ja-v1.0