Stt Fr Fastconformer Hybrid Large Pc

🚀 NVIDIA FastConformer-Hybrid Large (fr)

This model is designed for automatic speech recognition of French speech. It can transcribe speech in both upper and lower case French, including spaces, periods, commas, and question marks. It's a large - scale FastConformer Transducer - CTC model with around 115M parameters, trained on Transducer and CTC losses.

| | |

🚀 Quick Start

NVIDIA NeMo: Training

To train, fine - tune or play with the model, you need to install NVIDIA NeMo. It's recommended to install it after the latest Pytorch version.

pip install nemo_toolkit['all']

✨ Features

• Transcribes French speech with upper and lower case, along with basic punctuation.
• A hybrid model trained on Transducer and CTC losses.
• Based on the FastConformer architecture, an optimized version of Conformer.

📦 Installation

To use the model, first install the NeMo toolkit as described above.

💻 Usage Examples

Automatically instantiate the model

import nemo.collections.asr as nemo_asr
asr_model = nemo_asr.models.EncDecHybridRNNTCTCBPEModel.from_pretrained(model_name="nvidia/stt_fr_fastconformer_hybrid_large_pc")

Transcribing using Python

First, get a sample:

wget https://dldata-public.s3.us-east-2.amazonaws.com/2086-149220-0033.wav

Then transcribe:

output = asr_model.transcribe(['2086-149220-0033.wav'])
print(output[0].text)

Transcribing many audio files

Using Transducer mode inference:

python [NEMO_GIT_FOLDER]/examples/asr/transcribe_speech.py 
 pretrained_name="nvidia/stt_fr_fastconformer_hybrid_large_pc" 
 audio_dir="<DIRECTORY CONTAINING AUDIO FILES>"

Using CTC mode inference:

python [NEMO_GIT_FOLDER]/examples/asr/transcribe_speech.py 
 pretrained_name="nvidia/stt_fr_fastconformer_hybrid_large_pc" 
 audio_dir="<DIRECTORY CONTAINING AUDIO FILES>"
 decoder_type="ctc"

Input

This model accepts 16000 Hz Mono - channel Audio (wav files) as input.

Output

This model provides transcribed speech as a string for a given audio sample.

📚 Documentation

Model Architecture

FastConformer [1] is an optimized version of the Conformer model with 8x depthwise - separable convolutional downsampling. The model is trained in a multitask setup with joint Transducer and CTC decoder loss. You can find more information on the details of FastConformer here: Fast - Conformer Model and about Hybrid Transducer - CTC training here: Hybrid Transducer - CTC.

Training

The NeMo toolkit [3] was used for training the models for over several hundred epochs. These models are trained with this example script and this base config.

The tokenizers for these models were built using the text transcripts of the train set with this script.

Datasets

All the models in this collection are trained on a composite dataset (NeMo PnC ASRSET) comprising of 1800 hours of French speech:

• MCV12 (710 hrs)
• MLS (925 hrs)
• Voxpopuli (165 hrs)

Performance

The performance of Automatic Speech Recognition models is measured using Word Error Rate. Since this dataset is trained on multiple domains and a much larger corpus, it will generally perform better at transcribing audio in general.

The following tables summarize the performance of the available models in this collection with the Transducer decoder. Performances of the ASR models are reported in terms of Word Error Rate (WER%) with greedy decoding.

a) On data without Punctuation and Capitalization with Transducer decoder

b) On data with Punctuation and Capitalization with Transducer decoder

🔧 Technical Details

The model is a hybrid of Transducer and CTC losses, which combines the advantages of both. The FastConformer architecture provides efficient feature extraction with 8x depthwise - separable convolutional downsampling, enabling better performance in speech recognition tasks.

📄 License

License to use this model is covered by the CC - BY - 4.0. By downloading the public and release version of the model, you accept the terms and conditions of the CC - BY - 4.0 license.

Limitations

Since this model was trained on publicly available speech datasets, the performance of this model might degrade for speech which includes technical terms, or vernacular that the model has not been trained on. The model might also perform worse for accented speech. The model only outputs the punctuations: '.', ',', '?' and hence might not do well in scenarios where other punctuations are also expected.

NVIDIA Riva: Deployment

NVIDIA Riva is an accelerated speech AI SDK deployable on - prem, in all clouds, multi - cloud, hybrid, on edge, and embedded. Additionally, Riva provides:

• World - class out - of - the - box accuracy for the most common languages with model checkpoints trained on proprietary data with hundreds of thousands of GPU - compute hours
• Best in class accuracy with run - time word boosting (e.g., brand and product names) and customization of acoustic model, language model, and inverse text normalization
• Streaming speech recognition, Kubernetes compatible scaling, and enterprise - grade support

Although this model isn’t supported yet by Riva, the list of supported models is here.
Check out Riva live demo.

References

[1] Fast Conformer with Linearly Scalable Attention for Efficient Speech Recognition [2] Google Sentencepiece Tokenizer [3] NVIDIA NeMo Toolkit