Hao Li

Hawau Olamide Toyin, Hao Li, Hanan Aldarmaki

Speech recognition and speech synthesis models are typically trained separately, each with its own set of learning objectives, training data, and model parameters, resulting in two distinct large networks. We propose a parameter-efficient approach to learning ASR and TTS jointly via a multi-task learning objective and shared parameters. Our evaluation demonstrates that the performance of our multi-task model is comparable to that of individually trained models while significantly saving computational and memory costs ($\sim$50\% reduction in the total number of parameters required for the two tasks combined). We experiment with English as a resource-rich language, and Arabic as a relatively low-resource language due to shortage of TTS data. Our models are trained with publicly available data, and both the training code and model checkpoints are openly available for further research.

Kanghao Zhang, Shulin He, Hao Li et al.

In real acoustic environment, speech enhancement is an arduous task to improve the quality and intelligibility of speech interfered by background noise and reverberation. Over the past years, deep learning has shown great potential on speech enhancement. In this paper, we propose a novel real-time framework called DBNet which is a dual-branch structure with alternate interconnection. Each branch incorporates an encoder-decoder architecture with skip connections. The two branches are responsible for spectrum and waveform modeling, respectively. A bridge layer is adopted to exchange information between the two branches. Systematic evaluation and comparison show that the proposed system substantially outperforms related algorithms under very challenging environments. And in INTERSPEECH 2021 Deep Noise Suppression (DNS) challenge, the proposed system ranks the top 8 in real-time track 1 in terms of the Mean Opinion Score (MOS) of the ITU-T P.835 framework.

Hao Li, Xueliang Zhang, Guanglai Gao

Deep learning has shown a great potential for speech separation, especially for speech and non-speech separation. However, it encounters permutation problem for multi-speaker separation where both target and interference are speech. Permutation Invariant training (PIT) was proposed to solve this problem by permuting the order of the multiple speakers. Another way is to use an anchor speech, a short speech of the target speaker, to model the speaker identity. In this paper, we propose a simple strategy to train a long short-term memory (LSTM) model to solve the permutation problem in speaker separation. Specifically, we insert a short speech of target speaker at the beginning of a mixture as guide information. So, the first appearing speaker is defined as the target. Due to the powerful capability on sequence modeling, LSTM can use its memory cells to track and separate target speech from interfering speech. Experimental results show that the proposed training strategy is effective for speaker separation.

Shasha Xia, Hao Li, Xueliang Zhang

Supervised speech separation uses supervised learning algorithms to learn a mapping from an input noisy signal to an output target. With the fast development of deep learning, supervised separation has become the most important direction in speech separation area in recent years. For the supervised algorithm, training target has a significant impact on the performance. Ideal ratio mask is a commonly used training target, which can improve the speech intelligibility and quality of the separated speech. However, it does not take into account the correlation between noise and clean speech. In this paper, we use the optimal ratio mask as the training target of the deep neural network (DNN) for speech separation. The experiments are carried out under various noise environments and signal to noise ratio (SNR) conditions. The results show that the optimal ratio mask outperforms other training targets in general.