You can deploy this model and start earning money today!

FredrikKarlssonSpeech/
nb-whisper-small-onnx

Name: nb-whisper-small-onnx
Availability: InStock

transformers

onnx

whisper

automatic-speech-recognition

NbAiLab/nb-whisper-small

quantized

apache-2.0

endpoints_compatible

Views

No views yet

nb-whisper-small (ONNX)

ONNX conversion of NbAiLab/nb-whisper-small with multiple precision variants.

Available Precisions

Variant	Description	Files
fp32	Full precision (float32)	`encoder_model.onnx`, `decoder_model.onnx`, `decoder_with_past_model.onnx`, `decoder_model_merged.onnx`
fp16	Half precision (float16)	`*_fp16.onnx`
int8	8-bit integer (signed)	`*_int8.onnx`
uint8	8-bit integer (unsigned)	`*_uint8.onnx`
quantized	Dynamic quantization (avx512_vnni)	`*_quantized.onnx`
q4	4-bit weight quantization	`*_q4.onnx`
q4f16	4-bit weights + fp16 scales	`*_q4f16.onnx`
bnb4	BitsAndBytes NF4 quantization	`*_bnb4.onnx`

Usage with Optimum

python

from optimum.onnxruntime import ORTModelForSpeechSeq2Seq
from transformers import AutoProcessor
import librosa

model_id = "path/to/nb-whisper-small"
processor = AutoProcessor.from_pretrained(model_id)
model = ORTModelForSpeechSeq2Seq.from_pretrained(model_id, subfolder="onnx")

audio, sr = librosa.load("audio.wav", sr=16000, mono=True)

inputs = processor.feature_extractor(audio, sampling_rate=16000, return_tensors="pt")
gen_tokens = model.generate(**inputs, max_length=300)
print(processor.decode(gen_tokens[0], skip_special_tokens=True))

Usage with transformers.js

In a browser environment with AudioContext available:

javascript

import { pipeline } from "@huggingface/transformers";

const transcriber = await pipeline(
  "automatic-speech-recognition",
  "path/to/nb-whisper-small",
);
const result = await transcriber("audio.wav");
console.log(result.text);

In Node.js (AudioContext is not available, so audio must be loaded manually):

javascript

import { pipeline } from "@huggingface/transformers";
import fs from "fs";
import wavefile from "wavefile";

// Load and resample audio to 16kHz mono
const buffer = fs.readFileSync("audio.wav");
const wav = new wavefile.WaveFile(buffer);
wav.toBitDepth("32f");
wav.toSampleRate(16000);
const audioData = wav.getSamples(false, Float32Array);
// Mix to mono if stereo
const audio = Array.isArray(audioData)
  ? audioData[0].map((v, i) => (v + audioData[1][i]) / 2)
  : audioData;

const transcriber = await pipeline(
  "automatic-speech-recognition",
  "path/to/nb-whisper-small",
);
const result = await transcriber(audio);
console.log(result.text);

Original Model

This is an ONNX conversion of NbAiLab/nb-whisper-small. See the original model card for training details, performance metrics, and dataset information.

Language: Norwegian Bokmal
Task: Automatic Speech Recognition
License: Apache 2.0
Original authors: NbAiLab

Conversion Details

Exported with Hugging Face Optimum
Quantized with ONNX Runtime quantization tools
fp16 conversion via onnxconverter-common
4-bit quantization via MatMulNBitsQuantizer (q4) and MatMulBnb4Quantizer (bnb4)

FredrikKarlssonSpeech/nb-whisper-small-onnx