Model reference · open weights
higgs-audio-stt is an open-weight audio or speech model from bosonai. AxForge deploys and operates it for you on dedicated EU-owned hardware — with the licence handled where one is required.
Available as managed deployment — configured and operated for you on dedicated EU hardware, quoted per deployment.
What it is
| Released by | bosonai |
|---|---|
| Type | Audio & music |
| Task | Speech→text |
| Parameters (lead) | 8.9B |
| Context | 32k tokens |
| Based on | bosonai/higgs-audio-v3-8b |
| Released | 2026-04-27 |
| Popularity | 19k downloads / month |
| Licence | Open weights |
About
A speech-to-text model combining a Whisper-Large-v3 encoder with a Qwen3-8B decoder (8.91B total parameters), fine-tuned with LoRA on diverse ASR benchmarks.
import torch
import numpy as np
from transformers import AutoModel, AutoTokenizer
# Load model
model = AutoModel.from_pretrained(
"bosonai/higgs-audio-v3-8b-stt-v2",
torch_dtype=torch.bfloat16,
trust_remote_code=True,
attn_implementation="eager",
device_map="cuda:0",
)
tokenizer = AutoTokenizer.from_pretrained("bosonai/higgs-audio-v3-8b-stt-v2")
# Transcribe audio (16kHz mono numpy array)
from transformers.utils import cached_file
import importlib.util
spec = importlib.util.spec_from_file_location("transcribe", cached_file("bosonai/higgs-audio-v3-8b-stt-v2", "transcribe.py", _raise_exceptions_for_connection_errors=False))
mod = importlib.util.module_from_spec(spec)
spec.loader.exec_module(mod)
audio_np = np.random.randn(16000).astype(np.float32) # replace with your audio
text = mod.transcribe(model, tokenizer, audio_np)
print(text)
torch
transformers>=4.51.0
whisper # for audio preprocessing (WhisperProcessor)
| Dataset | WER |
|---|---|
| AMI | 10.14% |
| Earnings22 | 8.73% |
| GigaSpeech | 8.47% |
| LibriSpeech Clean | 1.25% |
| LibriSpeech Other | 2.38% |
| SPGISpeech | 3.60% |
| TED-LIUM | 3.09% |
| VoxPopuli | 5.92% |
| Average | 5.449% |
From the published model card. Full card on the HuggingFace links in the sidebar.
Benchmarks
As published on the model card — the maker's own numbers, not measured by AxForge.
| Task | Dataset | Metric | Score |
|---|---|---|---|
| Automatic Speech Recognition | AMI (Meetings test) | Test WER | 10.140 |
| Automatic Speech Recognition | Earnings-22 | Test WER | 8.730 |
| Automatic Speech Recognition | GigaSpeech | Test WER | 8.470 |
| Automatic Speech Recognition | LibriSpeech (clean) | Test WER | 1.250 |
| Automatic Speech Recognition | LibriSpeech (other) | Test WER | 2.380 |
| Automatic Speech Recognition | SPGI Speech | Test WER | 3.600 |
| Automatic Speech Recognition | tedlium-v3 | Test WER | 3.090 |
| Automatic Speech Recognition | Vox Populi | Test WER | 5.920 |
Using it via the API
Once AxForge deploys higgs-audio-stt for you, it answers on the OpenAI-compatible API — the same base URL and keys as every other model. (higgs-audio-stt below is illustrative; you get the exact model name on deployment.)
$ curl -sS https://api.axforge.ai/v1/audio/transcriptions \ -H "Authorization: Bearer $AXFORGE_API_KEY" \ -F model="higgs-audio-stt" -F file=@audio.mp3
Create an account — your API key is available in the console. 3M free tokens every 30 days with every new account.