Generative Speech Foundation Model Pretraining for High-Quality Speech Extraction and Restoration | IEEE Conference Publication | IEEE Xplore

Generative Speech Foundation Model Pretraining for High-Quality Speech Extraction and Restoration


Abstract:

This paper proposes a generative pretraining foundation model for high-quality speech restoration tasks. By directly operating on complex-valued short-time Fourier transf...Show More

Abstract:

This paper proposes a generative pretraining foundation model for high-quality speech restoration tasks. By directly operating on complex-valued short-time Fourier transform coefficients, our model does not rely on any vocoders for time-domain signal reconstruction. As a result, our model simplifies the synthesis process and removes the quality upper-bound introduced by any mel-spectrogram vocoder compared to prior work SpeechFlow. The proposed method is evaluated on multiple speech restoration tasks, including speech denoising, bandwidth extension, codec artifact removal, and target speaker extraction. In all scenarios, finetuning our pretrained model results in superior performance over strong baselines. Notably, in the target speaker extraction task, our model outperforms existing systems, including those leveraging SSL-pretrained encoders like WavLM. The code and the pretrained checkpoints are publicly available in the NVIDIA NeMo framework.
Date of Conference: 06-11 April 2025
Date Added to IEEE Xplore: 07 March 2025
ISBN Information:

ISSN Information:

Conference Location: Hyderabad, India

Contact IEEE to Subscribe

References

References is not available for this document.