metadata

language: en
datasets:
  - common_voice
  - mozilla-foundation/common_voice_6_0
metrics:
  - wer
  - cer
tags:
  - audio
  - automatic-speech-recognition
  - en
  - hf-asr-leaderboard
  - mozilla-foundation/common_voice_6_0
  - robust-speech-event
  - speech
  - xlsr-fine-tuning-week
license: apache-2.0
model-index:
  - name: XLSR Wav2Vec2 English by Jonatas Grosman
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: Common Voice en
          type: common_voice
          args: en
        metrics:
          - name: Test WER
            type: wer
            value: 19.06
          - name: Test CER
            type: cer
            value: 7.69
          - name: Test WER (+LM)
            type: wer
            value: 14.81
          - name: Test CER (+LM)
            type: cer
            value: 6.84
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: Robust Speech Event - Dev Data
          type: speech-recognition-community-v2/dev_data
          args: en
        metrics:
          - name: Dev WER
            type: wer
            value: 27.72
          - name: Dev CER
            type: cer
            value: 11.65
          - name: Dev WER (+LM)
            type: wer
            value: 20.85
          - name: Dev CER (+LM)
            type: cer
            value: 11.01
base_model:
  - jonatasgrosman/wav2vec2-large-xlsr-53-english

Disclaimer and Requirements

This model is a clone of jonatasgrosman/wav2vec2-large-xlsr-53-english compressed using ZipNN. Compressed losslessly to 88% its original size, ZipNN saved ~0.2GB in storage and potentially ~4PB in data transfer monthly.

Requirement

In order to use the model, ZipNN is necessary:

pip install zipnn

Use This Model

# Use a pipeline as a high-level helper
from transformers import pipeline
from zipnn import zipnn_hf

zipnn_hf()


pipe = pipeline("automatic-speech-recognition", model="royleibov/wav2vec2-large-xlsr-53-english-ZipNN-Compressed")

# Load model directly
from transformers import AutoProcessor, AutoModelForCTC
from zipnn import zipnn_hf

zipnn_hf()

processor = AutoProcessor.from_pretrained("royleibov/wav2vec2-large-xlsr-53-english-ZipNN-Compressed")
model = AutoModelForCTC.from_pretrained("royleibov/wav2vec2-large-xlsr-53-english-ZipNN-Compressed")

ZipNN

ZipNN also allows you to seemlessly save local disk space in your cache after the model is downloaded.

To compress the cached model, simply run:

python zipnn_compress_path.py safetensors --model royleibov/wav2vec2-large-xlsr-53-english-ZipNN-Compressed --hf_cache

The model will be decompressed automatically and safely as long as zipnn_hf() is added at the top of the file like in the example above.

To decompress manualy, simply run:

python zipnn_decompress_path.py --model royleibov/wav2vec2-large-xlsr-53-english-ZipNN-Compressed --hf_cache

Fine-tuned XLSR-53 large model for speech recognition in English

Fine-tuned facebook/wav2vec2-large-xlsr-53 on English using the train and validation splits of Common Voice 6.1. When using this model, make sure that your speech input is sampled at 16kHz.

This model has been fine-tuned thanks to the GPU credits generously given by the OVHcloud :)

The script used for training can be found here: https://github.com/jonatasgrosman/wav2vec2-sprint

Usage

The model can be used directly (without a language model) as follows...

Using the HuggingSound library:

from huggingsound import SpeechRecognitionModel
from zipnn import zipnn_hf

zipnn_hf()

model = SpeechRecognitionModel("royleibov/wav2vec2-large-xlsr-53-english-ZipNN-Compressed")
audio_paths = ["/path/to/file.mp3", "/path/to/another_file.wav"]

transcriptions = model.transcribe(audio_paths)

Writing your own inference script:

import torch
import librosa
from datasets import load_dataset
from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
from zipnn import zipnn_hf

zipnn_hf()

LANG_ID = "en"
MODEL_ID = "royleibov/wav2vec2-large-xlsr-53-english-ZipNN-Compressed"
SAMPLES = 10

test_dataset = load_dataset("common_voice", LANG_ID, split=f"test[:{SAMPLES}]")

processor = Wav2Vec2Processor.from_pretrained(MODEL_ID)
model = Wav2Vec2ForCTC.from_pretrained(MODEL_ID)

# Preprocessing the datasets.
# We need to read the audio files as arrays
def speech_file_to_array_fn(batch):
    speech_array, sampling_rate = librosa.load(batch["path"], sr=16_000)
    batch["speech"] = speech_array
    batch["sentence"] = batch["sentence"].upper()
    return batch

test_dataset = test_dataset.map(speech_file_to_array_fn)
inputs = processor(test_dataset["speech"], sampling_rate=16_000, return_tensors="pt", padding=True)

with torch.no_grad():
    logits = model(inputs.input_values, attention_mask=inputs.attention_mask).logits

predicted_ids = torch.argmax(logits, dim=-1)
predicted_sentences = processor.batch_decode(predicted_ids)

for i, predicted_sentence in enumerate(predicted_sentences):
    print("-" * 100)
    print("Reference:", test_dataset[i]["sentence"])
    print("Prediction:", predicted_sentence)

Reference	Prediction
"SHE'LL BE ALL RIGHT."	SHE'LL BE ALL RIGHT
SIX	SIX
"ALL'S WELL THAT ENDS WELL."	ALL AS WELL THAT ENDS WELL
DO YOU MEAN IT?	DO YOU MEAN IT
THE NEW PATCH IS LESS INVASIVE THAN THE OLD ONE, BUT STILL CAUSES REGRESSIONS.	THE NEW PATCH IS LESS INVASIVE THAN THE OLD ONE BUT STILL CAUSES REGRESSION
HOW IS MOZILLA GOING TO HANDLE AMBIGUITIES LIKE QUEUE AND CUE?	HOW IS MOSLILLAR GOING TO HANDLE ANDBEWOOTH HIS LIKE Q AND Q
"I GUESS YOU MUST THINK I'M KINDA BATTY."	RUSTIAN WASTIN PAN ONTE BATTLY
NO ONE NEAR THE REMOTE MACHINE YOU COULD RING?	NO ONE NEAR THE REMOTE MACHINE YOU COULD RING
SAUCE FOR THE GOOSE IS SAUCE FOR THE GANDER.	SAUCE FOR THE GUICE IS SAUCE FOR THE GONDER
GROVES STARTED WRITING SONGS WHEN SHE WAS FOUR YEARS OLD.	GRAFS STARTED WRITING SONGS WHEN SHE WAS FOUR YEARS OLD

Evaluation

To evaluate on mozilla-foundation/common_voice_6_0 with split test

python eval.py --model_id royleibov/wav2vec2-large-xlsr-53-english-ZipNN-Compressed --dataset mozilla-foundation/common_voice_6_0 --config en --split test

To evaluate on speech-recognition-community-v2/dev_data

python eval.py --model_id royleibov/wav2vec2-large-xlsr-53-english-ZipNN-Compressed --dataset speech-recognition-community-v2/dev_data --config en --split validation --chunk_length_s 5.0 --stride_length_s 1.0

Citation

If you want to cite this model you can use this:

@misc{grosman2021xlsr53-large-english,
  title={Fine-tuned {XLSR}-53 large model for speech recognition in {E}nglish},
  author={Grosman, Jonatas},
  howpublished={\url{https://huggingface.co/jonatasgrosman/wav2vec2-large-xlsr-53-english}},
  year={2021}
}