thezholdoshbekov/fastwave_best_bad_good

--- license: mit task_categories: - audio-to-audio language: - en tags: - audio - super-resolution - speech-enhancement features: sample_id: dtype: int32 _type: Value original: sampling_rate: 48000 _type: Audio augmented: sampling_rate: 24000 _type: Audio reconstructed: sampling_rate: 48000 _type: Audio configs: - config_name: default data_files: - split: train path: data/train-* dataset_info: features: - name: sample_id dtype: int32 - name: original dtype: audio: sampling_rate: 48000 - name: augmented dtype: audio: sampling_rate: 24000 - name: reconstructed dtype: audio: sampling_rate: 48000 splits: - name: train num_bytes: 12528758 num_examples: 5 download_size: 12532517 dataset_size: 12528758 --- # FastWave Audio Super-Resolution Samples | Column | Description | |--------|-------------| | **original** | Clean studio audio (48kHz) | | **augmented** | Degraded input — telephone/codec/narrowband (24kHz) | | **reconstructed** | Model output after super-resolution (48kHz) |