modelscope/ClearerVoice-Studio

Features

• Speech Processing Toolkits - A comprehensive toolkit for enhancing, isolating, and upsampling speech audio using deep learning.
• Speech Enhancement Models - Provides a comprehensive framework for removing background noise and improving speech quality through denoising and super-resolution.
• Speech Denoisers - Removes background noise and enhances audio quality to make speech recordings clearer.
• Multimodal Speech Extraction - Implements speech extraction guided by acoustic data, facial movements, or brain activity.
• Primary Speaker Isolation - Isolates a specific target voice from mixed audio signals using reference cues.
• Reference-Based Isolation - Isolates specific voices from mixed signals using reference speech, facial movements, or brain activity signals.
• Quality Thresholds - Measures speech processing effectiveness using signal-to-noise ratios and perceptual quality scores.
• Speech Separation Models - Isolates overlapping voices by capturing long-range dependencies and temporal patterns within the audio waveform.
• Target Sound Extraction - Isolates a specific target voice from mixed audio using acoustic, visual, or neural reference signals.
• Audio Super-Resolution - Reconstructs high-fidelity audio from low-sampling-rate signals using deep learning to restore high-frequency content.
• Audio Quality Evaluation Tools - Calculates signal-to-noise ratios and perceptual scores to quantify audio clarity and distortion.
• Audio Signal Fidelity Metrics - Quantifies audio clarity and processing effectiveness using objective fidelity metrics and perceptual scores.
• Noise Suppression Model Training - Provides training pipelines to produce custom models for noise suppression in speech recordings.
• Model Fine-Tuning - Adapts pre-trained speech models to specific datasets to improve denoising and separation performance.
• Speech Model Training - Provides specialized training infrastructure for speech enhancement, separation, and resolution models.
• Super-Resolution Training - Provides scripts to train and fine-tune models that upscale low-resolution speech to high-fidelity audio.
• Audio Clarity Metrics - Calculates objective metrics and perceptual scores to measure audio distortion and background noise.
• Speech Quality Metrics - Quantifies speech processing effectiveness using signal-to-noise ratios and perceptual quality scores.

Star history