nvidia/stt_kk_ru_fastconformer_hybrid_large

NVIDIA FastConformer-Hybrid Large (kk-ru)

| | |

This model transcribes speech in lower case Kazakh and Russian alphabet. It is a "large" version of FastConformer Transducer-CTC (around 115M parameters) model. This is a hybrid model trained on two losses: Token-and-Duration Transducer (default) and CTC. 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.EncDecHybridRNNTCTCBPEModel.from_pretrained(model_name="nvidia/stt_kk_ru_fastconformer_hybrid_large")

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:

asr_model.transcribe(['2086-149220-0033.wav'])

Transcribing many audio files

Using Transducer mode inference:

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

Using CTC mode inference:

python [NEMO_GIT_FOLDER]/examples/asr/transcribe_speech.py 
 pretrained_name="nvidia/stt_kk_ru_fastconformer_hybrid_large" 
 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.

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 Token-and-Duration Transducer and CTC decoder loss. You may 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 model 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

The model is trained on two composite datasets comprising of 1550 hours of Kazakh speech:

• MCV 17.0 Kazakh (1 hrs)
• Kazakh Speech Dataset (KSD) (416 hrs)
• Kazakh Speech Corpus 2 (KSC2) (1133 hrs)

and approximately 850 hrs of Russian speech:

• Golos (604 hrs)
• Sova (122 hrs)
• Dusha (102 hrs)
• MCV12 (19 hrs)

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 model with the Transducer decoder. Performances of the ASR models are reported in terms of Word Error Rate (WER%) with greedy decoding.

a) On Kazakh data

Version	Tokenizer	Vocabulary Size	MCV 17.0 TEST	KSD TEST	KSC2 TEST Read	KSC2 TEST Spontaneous
2.0.0	SentencePiece Unigram	1024	15.48	7.08	4.43	15.25

b) On Russian data

Version	Tokenizer	Vocabulary Size	MCV12 TEST	Sova TEST RuDevices	Sova TEST RuAudiobooksDevices	GOLOS TEST FARFIELD	GOLOS TEST CROWD	DUSHA TEST
2.0.0	SentencePiece Unigram	1024	6.29	19.83	4.41	5.98	2.46	5.93

Limitations

The model is non-streaming and outputs the speech as a string without capitalization and punctuation. 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.

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

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.