A PyTorch implementation of Conv-TasNet described in "TasNet: Surpassing Ideal Time-Frequency Masking for Speech Separation" with Permutation Invariant Training (PIT).
A PyTorch implementation of DNN-based source separation.
Implementation of "SpEx: Multi-Scale Time Domain Speaker Extraction Network".
Efficient Personalized Speech Enhancement through Self-Supervised Learning
Target speaker extraction — isolate any voice from a noisy recording using a short reference clip. Conv-TasNet separator and ECAPA-TDNN encoder, both trained from scratch.
远场目标说话人提取：Conv-TasNet + 6 路 ITD/ILD 空间特征早融合
Bachelor Final Year Project exploring real-time speech denoising using machine learning. Compares classical methods (SS, WF, MMSE-LSA) with 5 deep models on spectrogram data, highlighting Conv-TasNet’s effectiveness. Features dataset bucketing, OOM mitigation, and batch evaluation.
About Implementation and training of a deep neural network for speech denoising tasks.
Who spoke what and when: speech separation, speaker diarization, Hindi/Hinglish transcription and a local-LLM summary report for noisy, overlapping, code-switched audio. CS F407 (AI) project, BITS Pilani.
Two-speaker speech separation, 14.29 dB SI-SNR. Conv-TasNet + SC-CHM with input-adaptive slice fusion (<3K extra params).
5-class 복합 오디오 딥페이크 탐지 | Source Separation + Physics-based Features + LightGBM
To associate your repository with the conv-tasnet topic, visit your repo's landing page and select "manage topics."