This model transcribes speech in lower case English alphabet along with spaces and apostrophes.
It is a large version of Conformer-Transducer (around 120M parameters) model.
See the model architecture section and NeMo documentation for complete architecture details.
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']
The model is available for use in the NeMo toolkit , and can be used as a pre-trained checkpoint for inference or for fine-tuning on another dataset.
import nemo.collections.asr as nemo_asr asr_model = nemo_asr.models.EncDecRNNTBPEModel.from_pretrained("nvidia/stt_en_conformer_transducer_large")
First, let's get a sample
Then simply do:
python [NEMO_GIT_FOLDER]/examples/asr/transcribe_speech.py pretrained_name="nvidia/stt_en_conformer_transducer_large" audio_dir="<DIRECTORY CONTAINING AUDIO FILES>"
This model accepts 16000 KHz Mono-channel Audio (wav files) as input.
This model provides transcribed speech as a string for a given audio sample.
Conformer-Transducer model is an autoregressive variant of Conformer model  for Automatic Speech Recognition which uses Transducer loss/decoding instead of CTC Loss. You may find more info on the detail of this model here: Conformer-Transducer Model.
The tokenizers for these models were built using the text transcripts of the train set with this script.
All the models in this collection are trained on a composite dataset (NeMo ASRSET) comprising of several thousand hours of English speech:
- Librispeech 960 hours of English speech
- Fisher Corpus
- Switchboard-1 Dataset
- WSJ-0 and WSJ-1
- National Speech Corpus (Part 1, Part 6)
- VoxPopuli (EN)
- Europarl-ASR (EN)
- Multilingual Librispeech (MLS EN) - 2,000 hrs subset
- Mozilla Common Voice (v8.0)
- People's Speech - 12,000 hrs subset
Note: older versions of the model may have trained on smaller set of datasets.
The list of the available models in this collection is shown in the following table. Performances of the ASR models are reported in terms of Word Error Rate (WER%) with greedy decoding.
|Version||Tokenizer||Vocabulary Size||LS test-other||LS test-clean||WSJ Eval92||WSJ Dev93||NSC Part 1||MLS Test||MCV Test 6.1||MCV Test 8.0||Train Dataset|
|1.10.0||SentencePiece Unigram||1024||3.7||1.7||1.5||2.1||5.9||5.8||6.5||7.8||NeMo ASRSET 3.0|
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.
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
- Downloads last month
Datasets used to train nvidia/stt_en_conformer_transducer_large
Space using nvidia/stt_en_conformer_transducer_large 1
- Test WER on LibriSpeech (clean)test set self-reported1.700
- Test WER on LibriSpeech (other)test set self-reported3.700
- Test WER on Multilingual LibriSpeechtest set self-reported5.800
- Test WER on Mozilla Common Voice 8.0test set self-reported7.800
- Test WER on Wall Street Journal 92self-reported1.500
- Test WER on Wall Street Journal 93self-reported2.100
- Test WER on National Singapore Corpusself-reported5.900