language:

• multilingual
• fr
• de
• es
• ca
• it
• ru
• zh
• pt
• fa
• et
• mn
• nl
• tr
• ar
• sv
• lv
• sl
• ta
• ja
• id
• cy
• en datasets:
• common_voice
• multilingual_librispeech
• covost2 tags:
• speech
• xls_r
• automatic-speech-recognition
• xls_r_translation pipeline_tag: automatic-speech-recognition license: apache-2.0 widget:
• example_title: Swedish src: https://cdn-media.huggingface.co/speech_samples/cv_swedish_1.mp3
• example_title: Arabic src: https://cdn-media.huggingface.co/speech_samples/common_voice_ar_19058308.mp3
• example_title: Russian src: https://cdn-media.huggingface.co/speech_samples/common_voice_ru_18849022.mp3
• example_title: German src: https://cdn-media.huggingface.co/speech_samples/common_voice_de_17284683.mp3
• example_title: French src: https://cdn-media.huggingface.co/speech_samples/common_voice_fr_17299386.mp3
• example_title: Indonesian src: https://cdn-media.huggingface.co/speech_samples/common_voice_id_19051309.mp3
• example_title: Italian src: https://cdn-media.huggingface.co/speech_samples/common_voice_it_17415776.mp3
• example_title: Japanese src: https://cdn-media.huggingface.co/speech_samples/common_voice_ja_19482488.mp3
• example_title: Mongolian src: https://cdn-media.huggingface.co/speech_samples/common_voice_mn_18565396.mp3
• example_title: Dutch src: https://cdn-media.huggingface.co/speech_samples/common_voice_nl_17691471.mp3
• example_title: Russian src: https://cdn-media.huggingface.co/speech_samples/common_voice_ru_18849022.mp3
• example_title: Turkish src: https://cdn-media.huggingface.co/speech_samples/common_voice_tr_17341280.mp3
• example_title: Catalan src: https://cdn-media.huggingface.co/speech_samples/common_voice_ca_17367522.mp3
• example_title: English src: https://cdn-media.huggingface.co/speech_samples/common_voice_en_18301577.mp3
• example_title: Dutch src: https://cdn-media.huggingface.co/speech_samples/common_voice_nl_17691471.mp3

Wav2Vec2-XLS-R-2b-21-EN

Facebook's Wav2Vec2 XLS-R fine-tuned for Speech Translation.

This is a SpeechEncoderDecoderModel model. The encoder was warm-started from the facebook/wav2vec2-xls-r-1b checkpoint and the decoder from the facebook/mbart-large-50 checkpoint. Consequently, the encoder-decoder model was fine-tuned on 21 {lang} -> en translation pairs of the Covost2 dataset.

The model can translate from the following spoken languages {lang} -> en (English):

{fr, de, es, ca, it, ru, zh-CN, pt, fa, et, mn, nl, tr, ar, sv-SE, lv, sl, ta, ja, id, cy} -> en

For more information, please refer to Section 5.1.2 of the official XLS-R paper.

Usage

Demo

The model can be tested directly on the speech recognition widget on this model card! Simple record some audio in one of the possible spoken languages or pick an example audio file to see how well the checkpoint can translate the input.

Example

As this a standard sequence to sequence transformer model, you can use the generate method to generate the transcripts by passing the speech features to the model.

You can use the model directly via the ASR pipeline

from datasets import load_dataset
from transformers import pipeline

# replace following lines to load an audio file of your choice
librispeech_en = load_dataset("patrickvonplaten/librispeech_asr_dummy", "clean", split="validation")
audio_file = librispeech_en[0]["file"]

asr = pipeline("automatic-speech-recognition", model="facebook/wav2vec2-xls-r-1b-21-to-en", feature_extractor="facebook/wav2vec2-xls-r-1b-21-to-en")

translation = asr(audio_file)

or step-by-step as follows:

import torch
from transformers import Speech2Text2Processor, SpeechEncoderDecoderModel
from datasets import load_dataset

model = SpeechEncoderDecoderModel.from_pretrained("facebook/wav2vec2-xls-r-1b-21-to-en")
processor = Speech2Text2Processor.from_pretrained("facebook/wav2vec2-xls-r-1b-21-to-en")

ds = load_dataset("patrickvonplaten/librispeech_asr_dummy", "clean", split="validation")

inputs = processor(ds[0]["audio"]["array"], sampling_rate=ds[0]["audio"]["array"]["sampling_rate"], return_tensors="pt")
generated_ids = model.generate(input_ids=inputs["input_features"], attention_mask=inputs["attention_mask"])
transcription = processor.batch_decode(generated_ids)

Results {lang} -> en

See the row of XLS-R (1B) for the performance on Covost2 for this model.

More XLS-R models for {lang} -> en Speech Translation

• Wav2Vec2-XLS-R-300M-21-EN
• Wav2Vec2-XLS-R-1B-21-EN
• Wav2Vec2-XLS-R-2B-21-EN
• Wav2Vec2-XLS-R-2B-22-16