o ei@sPdZddlZddlZddlmZddlmZGdddeZGdddeZdS) aP Specifies the inference interfaces for speech enhancement modules. Authors: * Aku Rouhe 2021 * Peter Plantinga 2021 * Loren Lugosch 2020 * Mirco Ravanelli 2020 * Titouan Parcollet 2021 * Abdel Heba 2021 * Andreas Nautsch 2022, 2023 * Pooneh Mousavi 2023 * Sylvain de Langen 2023 * Adel Moumen 2023 * Pradnya Kandarkar 2023 N) Pretrained)lengths_arg_existsc@s:eZdZdZgdZdgZddZd ddZd d d ZdS) SpectralMaskEnhancementaYA ready-to-use model for speech enhancement. Arguments --------- See ``Pretrained``. Example ------- >>> import torch >>> from speechbrain.inference.enhancement import SpectralMaskEnhancement >>> # Model is downloaded from the speechbrain HuggingFace repo >>> tmpdir = getfixture("tmpdir") >>> enhancer = SpectralMaskEnhancement.from_hparams( ... source="speechbrain/metricgan-plus-voicebank", ... savedir=tmpdir, ... ) >>> enhanced = enhancer.enhance_file( ... "speechbrain/metricgan-plus-voicebank/example.wav" ... ) ) compute_stftspectral_magnituderesynth enhance_modelcCs"|j|}|j|}t|S)a3Compute the log spectral magnitude features for masking. Arguments --------- wavs : torch.Tensor A batch of waveforms to convert to log spectral mags. Returns ------- feats : torch.Tensor The log spectral magnitude features. )hparamsrrtorchlog1p)selfwavsfeatsr_/home/ubuntu/transcripts/venv/lib/python3.10/site-packages/speechbrain/inference/enhancement.pycompute_features1s  z(SpectralMaskEnhancement.compute_featuresNcCs\||j}||}|dur|jj||d}n|j|}t||}|jt ||S)aEnhance a batch of noisy waveforms. Arguments --------- noisy : torch.Tensor A batch of waveforms to perform enhancement on. lengths : torch.Tensor The lengths of the waveforms if the enhancement model handles them. Returns ------- wavs : torch.Tensor A batch of enhanced waveforms of the same shape as input. Nlengths) todevicermodsrr mulr rexpm1)r noisyrnoisy_featuresmaskenhancedrrr enhance_batchBs    z%SpectralMaskEnhancement.enhance_batchcKs||j|fi|}||j}|d}t|jr%|j|tdgd}n||}|dur9tj |||j j j d| dS)aEnhance a wav file. Arguments --------- filename : str Location on disk to load file for enhancement. output_filename : str If provided, writes enhanced data to this file. **kwargs : dict Arguments forwarded to ``load_audio``. Returns ------- wav : torch.Tensor The enhanced waveform. rg?rNurisrc sample_rate) load_audiorr unsqueezerrr tensor torchaudiosaver rr!squeezer filenameoutput_filenamekwargsrbatchrrrr enhance_file^s     z$SpectralMaskEnhancement.enhance_fileN) __name__ __module__ __qualname____doc__HPARAMS_NEEDEDMODULES_NEEDEDrrr-rrrrrs rc@s4eZdZdZdgZd ddZd ddZd dd ZdS) WaveformEnhancementa2A ready-to-use model for speech enhancement. Arguments --------- See ``Pretrained``. Example ------- >>> from speechbrain.inference.enhancement import WaveformEnhancement >>> # Model is downloaded from the speechbrain HuggingFace repo >>> tmpdir = getfixture("tmpdir") >>> enhancer = WaveformEnhancement.from_hparams( ... source="speechbrain/mtl-mimic-voicebank", ... savedir=tmpdir, ... ) >>> enhanced = enhancer.enhance_file( ... "speechbrain/mtl-mimic-voicebank/example.wav" ... ) rNcCs ||j}|j|\}}|S)aEnhance a batch of noisy waveforms. Arguments --------- noisy : torch.Tensor A batch of waveforms to perform enhancement on. lengths : torch.Tensor The lengths of the waveforms if the enhancement model handles them. Returns ------- torch.Tensor A batch of enhanced waveforms of the same shape as input. )rrrr)r rr enhanced_wav_rrrrs z!WaveformEnhancement.enhance_batchcKsL|j|fi|}|d}||}|dur!tj|||jjd|dS)aEnhance a wav file. Arguments --------- filename : str Location on disk to load file for enhancement. output_filename : str If provided, writes enhanced data to this file. **kwargs : dict Arguments forwarded to ``load_audio`` Returns ------- enhanced : torch.Tensor The enhanced waveform. rNr)r"r#rr%r&audio_normalizerr!r'r(rrrr-s   z WaveformEnhancement.enhance_filecCs |||S)z#Runs enhancement on the noisy input)r)r rrrrrforwards zWaveformEnhancement.forwardr.)r/r0r1r2r4rr-r9rrrrr5s    r5) r2r r% speechbrain.inference.interfacesrspeechbrain.utils.callchainsrrr5rrrrs  k