bofenghuang's picture
up
0627400
metadata
license: apache-2.0
language: fr
library_name: transformers
thumbnail: null
tags:
  - automatic-speech-recognition
  - hf-asr-leaderboard
  - whisper-event
datasets:
  - mozilla-foundation/common_voice_11_0
  - facebook/multilingual_librispeech
  - facebook/voxpopuli
  - google/fleurs
  - gigant/african_accented_french
metrics:
  - wer
model-index:
  - name: Fine-tuned whisper-medium model for ASR in French
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: Common Voice 11.0
          type: mozilla-foundation/common_voice_11_0
          config: fr
          split: test
          args: fr
        metrics:
          - name: WER (Greedy)
            type: wer
            value: 9.03
          - name: WER (Beam 5)
            type: wer
            value: 8.73
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: Multilingual LibriSpeech (MLS)
          type: facebook/multilingual_librispeech
          config: french
          split: test
          args: french
        metrics:
          - name: WER (Greedy)
            type: wer
            value: 4.6
          - name: WER (Beam 5)
            type: wer
            value: 4.44
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: VoxPopuli
          type: facebook/voxpopuli
          config: fr
          split: test
          args: fr
        metrics:
          - name: WER (Greedy)
            type: wer
            value: 9.53
          - name: WER (Beam 5)
            type: wer
            value: 9.46
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: Fleurs
          type: google/fleurs
          config: fr_fr
          split: test
          args: fr_fr
        metrics:
          - name: WER (Greedy)
            type: wer
            value: 6.33
          - name: WER (Beam 5)
            type: wer
            value: 5.94
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: African Accented French
          type: gigant/african_accented_french
          config: fr
          split: test
          args: fr
        metrics:
          - name: WER (Greedy)
            type: wer
            value: 4.89
          - name: WER (Beam 5)
            type: wer
            value: 4.56

Model architecture Model size Language

Fine-tuned whisper-medium model for ASR in French

This model is a fine-tuned version of openai/whisper-medium, trained on a composite dataset comprising of over 2200 hours of French speech audio, using the train and the validation splits of Common Voice 11.0, Multilingual LibriSpeech, Voxpopuli, Fleurs, Multilingual TEDx, MediaSpeech, and African Accented French. When using the model make sure that your speech input is sampled at 16Khz. This model doesn't predict casing or punctuation.

Performance

Below are the WERs of the pre-trained models on the Common Voice 9.0, Multilingual LibriSpeech, Voxpopuli and Fleurs. These results are reported in the original paper.

Model Common Voice 9.0 MLS VoxPopuli Fleurs
openai/whisper-small 22.7 16.2 15.7 15.0
openai/whisper-medium 16.0 8.9 12.2 8.7
openai/whisper-large 14.7 8.9 11.0 7.7
openai/whisper-large-v2 13.9 7.3 11.4 8.3

Below are the WERs of the fine-tuned models on the Common Voice 11.0, Multilingual LibriSpeech, Voxpopuli, and Fleurs. Note that these evaluation datasets have been filtered and preprocessed to only contain French alphabet characters and are removed of punctuation outside of apostrophe. The results in the table are reported as WER (greedy search) / WER (beam search with beam width 5).

Model Common Voice 11.0 MLS VoxPopuli Fleurs
bofenghuang/whisper-small-cv11-french 11.76 / 10.99 9.65 / 8.91 14.45 / 13.66 10.76 / 9.83
bofenghuang/whisper-medium-cv11-french 9.03 / 8.54 6.34 / 5.86 11.64 / 11.35 7.13 / 6.85
bofenghuang/whisper-medium-french 9.03 / 8.73 4.60 / 4.44 9.53 / 9.46 6.33 / 5.94
bofenghuang/whisper-large-v2-cv11-french 8.05 / 7.67 5.56 / 5.28 11.50 / 10.69 5.42 / 5.05
bofenghuang/whisper-large-v2-french 8.15 / 7.83 4.20 / 4.03 9.10 / 8.66 5.22 / 4.98

Usage

Inference with 🤗 Pipeline

import torch

from datasets import load_dataset
from transformers import pipeline

device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")

# Load pipeline
pipe = pipeline("automatic-speech-recognition", model="bofenghuang/whisper-medium-french", device=device)

# NB: set forced_decoder_ids for generation utils
pipe.model.config.forced_decoder_ids = pipe.tokenizer.get_decoder_prompt_ids(language="fr", task="transcribe")

# Load data
ds_mcv_test = load_dataset("mozilla-foundation/common_voice_11_0", "fr", split="test", streaming=True)
test_segment = next(iter(ds_mcv_test))
waveform = test_segment["audio"]

# Run
generated_sentences = pipe(waveform, max_new_tokens=225)["text"]  # greedy
# generated_sentences = pipe(waveform, max_new_tokens=225, generate_kwargs={"num_beams": 5})["text"]  # beam search

# Normalise predicted sentences if necessary

Inference with 🤗 low-level APIs

import torch
import torchaudio

from datasets import load_dataset
from transformers import AutoProcessor, AutoModelForSpeechSeq2Seq

device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")

# Load model
model = AutoModelForSpeechSeq2Seq.from_pretrained("bofenghuang/whisper-medium-french").to(device)
processor = AutoProcessor.from_pretrained("bofenghuang/whisper-medium-french", language="french", task="transcribe")

# NB: set forced_decoder_ids for generation utils
model.config.forced_decoder_ids = processor.get_decoder_prompt_ids(language="fr", task="transcribe")

# 16_000
model_sample_rate = processor.feature_extractor.sampling_rate

# Load data
ds_mcv_test = load_dataset("mozilla-foundation/common_voice_11_0", "fr", split="test", streaming=True)
test_segment = next(iter(ds_mcv_test))
waveform = torch.from_numpy(test_segment["audio"]["array"])
sample_rate = test_segment["audio"]["sampling_rate"]

# Resample
if sample_rate != model_sample_rate:
    resampler = torchaudio.transforms.Resample(sample_rate, model_sample_rate)
    waveform = resampler(waveform)

# Get feat
inputs = processor(waveform, sampling_rate=model_sample_rate, return_tensors="pt")
input_features = inputs.input_features
input_features = input_features.to(device)

# Generate
generated_ids = model.generate(inputs=input_features, max_new_tokens=225)  # greedy
# generated_ids = model.generate(inputs=input_features, max_new_tokens=225, num_beams=5)  # beam search

# Detokenize
generated_sentences = processor.batch_decode(generated_ids, skip_special_tokens=True)[0]

# Normalise predicted sentences if necessary