/ infer.pyLatest commit Cannot retrieve latest commit at this time. HistoryHistory File metadata and controls Code Blame 24 lines (19 loc) · 782 Bytes Raw import os import torch import soundfile as sf from gtcrn import GTCRN ## load model device = torch.device("cpu") model = GTCRN(...
测试集中的两个典型示例在图5中展示,清楚地显示了GTCRN比RNNoise具有更优的噪声抑制效果。 源代码和音频示例:https://github.com/Xiaobin-Rong/gtcrn。 四、CONCLUSION 本文介绍了GTCRN,这是一个语音增强模型,仅需要23.7K参数和每秒39.6 MMACs。对DPCRN应用了多种策略,有效地减小了模型规模,同时保持了语音增强性...
stream modules onnx_models test_wavs README.md gtcrn.py gtcrn_stream.py test_wavs LICENSE README.md gtcrn.py infer.py loss.py requirements.txt Streaming GTCRN The streaming conversion method used here is consistent with the streaming conversion in theTRT-SErepository. However, we only depl...