NVIDIA Conformer-Transducer Large (Esperanto)

| | |

This model transcribes speech into lowercase Esperanto alphabet including spaces and apostroph. The model was obtained by finetuning from English SSL-pretrained model on Mozilla Common Voice Esperanto 11.0 dataset. It is a non-autoregressive "large" variant of Conformer [1], with around 120 million parameters. See the model architecture section and NeMo documentation for complete architecture details.

Usage

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

To train, finetune 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']

Automatically instantiate the model

import nemo.collections.asr as nemo_asr
asr_model = nemo_asr.models.EncDecRNNTBPEModel.from_pretrained("nvidia/stt_eo_conformer_transducer_large")

Transcribing using Python

Simply do:

asr_model.transcribe(['<your_audio>.wav'])

Transcribing many audio files

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

Input

This model accepts 16 kHz mono-channel Audio (wav files) as input.

Output

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

Model Architecture

Conformer-Transducer model is an autoregressive variant of Conformer model [1] for Automatic Speech Recognition which uses Transducer loss/decoding. You may find more info on the detail of this model here: Conformer-Transducer Model.

Training

The NeMo toolkit [3] was used for finetuning from English SSL model for over several hundred epochs. The model is finetuning with this example script and this base config. As pretrained English SSL model we use ssl_en_conformer_large which was trained using LibriLight corpus (~56k hrs of unlabeled English speech).

The tokenizer for the model was built using the text transcripts of the train set with this script.

Full config can be found inside the .nemo files.

Datasets

All the models were trained on Mozilla Common Voice Esperanto 11.0 dataset comprising of about 1400 validated hours of Esperanto speech. However, training set consists of a much smaller amount of data, because when forming the train.tsv, dev.tsv and test.tsv, repetitions of texts in train were removed by Mozilla developers.

Train set: ~250 hours.
Dev set: ~25 hours.
Test: ~25 hours.

Performance

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	Dev WER	Test WER	Train Dataset
1.14.0	SentencePiece [2] BPE	128	2.4	4.0	MCV-11.0 Train set

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.

Deployment with NVIDIA Riva

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.