language:

• en library_name: nemo datasets:
• librispeech_asr
• fisher_corpus
• mozilla-foundation/common_voice_8_0
• National-Singapore-Corpus-Part-1
• vctk
• voxpopuli
• europarl
• multilingual_librispeech

thumbnail: null tags:

• automatic-speech-recognition
• speech
• audio
• Transducer
• TDT
• FastConformer
• Conformer
• pytorch
• NeMo
• hf-asr-leaderboard license: cc-by-4.0 widget:
• example_title: Librispeech sample 1 src: https://cdn-media.huggingface.co/speech_samples/sample1.flac
• example_title: Librispeech sample 2 src: https://cdn-media.huggingface.co/speech_samples/sample2.flac model-index:
• name: parakeet-tdt_ctc-1.1b results:
  • task: name: Automatic Speech Recognition type: automatic-speech-recognition dataset: name: AMI (Meetings test) type: edinburghcstr/ami config: ihm split: test args: language: en metrics:
    • name: Test WER type: wer value: 15.94
  • task: name: Automatic Speech Recognition type: automatic-speech-recognition dataset: name: Earnings-22 type: revdotcom/earnings22 split: test args: language: en metrics:
    • name: Test WER type: wer value: 11.86
  • task: name: Automatic Speech Recognition type: automatic-speech-recognition dataset: name: GigaSpeech type: speechcolab/gigaspeech split: test args: language: en metrics:
    • name: Test WER type: wer value: 10.19
  • task: name: Automatic Speech Recognition type: automatic-speech-recognition dataset: name: LibriSpeech (clean) type: librispeech_asr config: other split: test args: language: en metrics:
    • name: Test WER type: wer value: 1.82
  • task: name: Automatic Speech Recognition type: automatic-speech-recognition dataset: name: LibriSpeech (other) type: librispeech_asr config: other split: test args: language: en metrics:
    • name: Test WER type: wer value: 3.67
  • task: type: Automatic Speech Recognition name: automatic-speech-recognition dataset: name: SPGI Speech type: kensho/spgispeech config: test split: test args: language: en metrics:
    • name: Test WER type: wer value: 2.24
  • task: type: Automatic Speech Recognition name: automatic-speech-recognition dataset: name: tedlium-v3 type: LIUM/tedlium config: release1 split: test args: language: en metrics:
    • name: Test WER type: wer value: 3.87
  • task: name: Automatic Speech Recognition type: automatic-speech-recognition dataset: name: Vox Populi type: facebook/voxpopuli config: en split: test args: language: en metrics:
    • name: Test WER type: wer value: 6.19
  • task: type: Automatic Speech Recognition name: automatic-speech-recognition dataset: name: Mozilla Common Voice 9.0 type: mozilla-foundation/common_voice_9_0 config: en split: test args: language: en metrics:
    • name: Test WER type: wer value: 8.69

metrics:

• wer pipeline_tag: automatic-speech-recognition

Parakeet TDT-CTC 1.1B PnC(en)

| |

parakeet-tdt_ctc-1.1b is an ASR model that transcribes speech with Punctuations and Capitalizations of English alphabet. This model is jointly developed by NVIDIA NeMo and Suno.ai teams. It is an XXL version of Hybrid FastConformer [1] TDT-CTC [2] (around 1.1B parameters) model. This model has been trained with Local Attention and Global token hence this model can transcribe 11 hrs of audio in one single pass. And for reference this model can transcibe 90mins of audio in <16 sec on A100.
See the model architecture section and NeMo documentation for complete architecture details.

NVIDIA NeMo: Training

To train, fine-tune or play with the model you will need to install NVIDIA NeMo. We recommend you install it after you've installed latest PyTorch version.

pip install nemo_toolkit['all']

How to Use this Model

The model is available for use in the NeMo toolkit [3], and can be used as a pre-trained checkpoint for inference or for fine-tuning on another dataset.

Automatically instantiate the model

import nemo.collections.asr as nemo_asr
asr_model = nemo_asr.models.ASRModel.from_pretrained(model_name="nvidia/parakeet-tdt_ctc-1.1b")

Transcribing using Python

First, let's get a sample

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

Then simply do:

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

Transcribing many audio files

By default model uses TDT to transcribe the audio files, to switch decoder to use CTC, use decoding_type='ctc'

python [NEMO_GIT_FOLDER]/examples/asr/transcribe_speech.py 
 pretrained_name="nvidia/parakeet-tdt_ctc-1.1b" 
 audio_dir="<DIRECTORY CONTAINING AUDIO FILES>"

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.

Model Architecture

This model uses a Hybrid FastConformer-TDT-CTC architecture. FastConformer [1] is an optimized version of the Conformer model with 8x depthwise-separable convolutional downsampling. You may find more information on the details of FastConformer here: Fast-Conformer Model.

Training

The NeMo toolkit [3] was used for finetuning this model for 20,000 steps over parakeet-tdt-1.1 model. This model is 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

The model was trained on 36K hours of English speech collected and prepared by NVIDIA NeMo and Suno teams.

The training dataset consists of private subset with 27K hours of English speech plus 9k hours from the following public PnC datasets:

• Librispeech 960 hours of English speech
• Fisher Corpus
• National Speech Corpus Part 1
• VCTK
• VoxPopuli (EN)
• Europarl-ASR (EN)
• Multilingual Librispeech (MLS EN) - 2,000 hour subset
• Mozilla Common Voice (v7.0)

Performance

The performance of Automatic Speech Recognition models is measuring 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 summarizes 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.

These are greedy WER numbers without external LM. More details on evaluation can be found at HuggingFace ASR Leaderboard

Model Fairness Evaluation

As outlined in the paper "Towards Measuring Fairness in AI: the Casual Conversations Dataset", we assessed the parakeet-tdt_ctc-1.1b model for fairness. The model was evaluated on the CausalConversations-v1 dataset, and the results are reported as follows:

Gender Bias:

Age Bias:

(Error rates for fairness evaluation are determined by normalizing both the reference and predicted text, similar to the methods used in the evaluations found at https://github.com/huggingface/open_asr_leaderboard.)

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] Efficient Sequence Transduction by Jointly Predicting Tokens and Durations

[3] Google Sentencepiece Tokenizer

[4] NVIDIA NeMo Toolkit

[5] Suno.ai

[6] HuggingFace ASR Leaderboard

[7] Towards Measuring Fairness in AI: the Casual Conversations Dataset

Licence

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.