Audio-to-Audio
MLX
Safetensors
speech-enhancement
speech-denoising
gtcrn
apple-silicon
onnx-conversion
Instructions to use agentable/gtcrn-simple-mlx with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use agentable/gtcrn-simple-mlx with MLX:
# Download the model from the Hub pip install huggingface_hub[hf_xet] huggingface-cli download --local-dir gtcrn-simple-mlx agentable/gtcrn-simple-mlx
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
- Atomic Chat
| { | |
| "protocol": "gtcrn-simple-v1", | |
| "model_type": "gtcrn", | |
| "sample_rate": 16000, | |
| "resample": { | |
| "kind": "sherpa-linear-resample", | |
| "lowpass_cutoff_ratio": 0.99, | |
| "lowpass_filter_width": 6, | |
| "flush": true | |
| }, | |
| "stft": { | |
| "n_fft": 512, | |
| "hop_length": 256, | |
| "window_length": 512, | |
| "window_type": "hann_sqrt", | |
| "bins": 257 | |
| }, | |
| "input": { | |
| "name": "mix", | |
| "dtype": "float32", | |
| "shape": [1, 257, 1, 2], | |
| "complex_layout": "real_imag" | |
| }, | |
| "caches": [ | |
| { | |
| "input_name": "conv_cache", | |
| "output_name": "conv_cache_out", | |
| "dtype": "float32", | |
| "shape": [2, 1, 16, 16, 33] | |
| }, | |
| { | |
| "input_name": "tra_cache", | |
| "output_name": "tra_cache_out", | |
| "dtype": "float32", | |
| "shape": [2, 3, 1, 1, 16] | |
| }, | |
| { | |
| "input_name": "inter_cache", | |
| "output_name": "inter_cache_out", | |
| "dtype": "float32", | |
| "shape": [2, 1, 33, 16] | |
| } | |
| ], | |
| "output": { | |
| "name": "enh", | |
| "dtype": "float32", | |
| "shape": [1, 257, 1, 2], | |
| "complex_layout": "real_imag" | |
| }, | |
| "limits": { | |
| "max_output_samples": 4800000 | |
| } | |
| } | |