o ۷i @sddlZddlmZmZddlmZz ddlZdZdZ Wne y6Z zdZde d Z WYdZ [ ndZ [ wwdd l m Z Gd d d eeZdS) N) ConfigMixinregister_to_config)SchedulerMixinTFzCannot import librosa because zB. Make sure to correctly install librosa to be able to install it.)Imagec@seZdZdZdZe       d'd ed ed ed ed ededefddZd ed efddZd(de de j fddZ defddZ d)dede j fddZdefd d!Zdedejfd"d#Zd$ejde j fd%d&ZdS)*Mela+ Parameters: x_res (`int`): x resolution of spectrogram (time). y_res (`int`): y resolution of spectrogram (frequency bins). sample_rate (`int`): Sample rate of audio. n_fft (`int`): Number of Fast Fourier Transforms. hop_length (`int`): Hop length (a higher number is recommended if `y_res` < 256). top_db (`int`): Loudest decibel value. n_iter (`int`): Number of iterations for Griffin-Lim Mel inversion. zmel_config.json"VP x_resy_res sample_raten_fft hop_lengthtop_dbn_itercCs@||_||_||_||_||_|||d|_tstt dS)N) rsrrrrset_resolutionaudio_librosa_can_be_imported ValueError _import_error)selfrrrrrrrrh/home/ubuntu/vllm_env/lib/python3.10/site-packages/diffusers/pipelines/deprecated/audio_diffusion/mel.py__init__:s  z Mel.__init__cCs*||_||_|j|_|j|jd|_dS)zSet resolution. Args: x_res (`int`): x resolution of spectrogram (time). y_res (`int`): y resolution of spectrogram (frequency bins). N)rrn_melsr slice_size)rrrrrrrPs zMel.set_resolutionN audio_file raw_audiocCsr|durtj|d|jd\|_}n||_t|j|j|jkr7t|jt |j|jt|jfg|_dSdS)aLoad audio. Args: audio_file (`str`): An audio file that must be on disk due to [Librosa](https://librosa.org/) limitation. raw_audio (`np.ndarray`): The raw audio file as a NumPy array. NT)monor) librosaloadrrlenrrnp concatenatezeros)rr#r$_rrr load_audio^s  0zMel.load_audioreturncCst|j|jS)zGet number of slices in audio. Returns: `int`: Number of spectograms audio can be sliced into. )r(rr"rrrrget_number_of_slicespszMel.get_number_of_slicesrslicecCs|j|j||j|dS)zGet slice of audio. Args: slice (`int`): Slice number of audio (out of `get_number_of_slices()`). Returns: `np.ndarray`: The audio slice as a NumPy array. r )rr")rr1rrrget_audio_sliceys zMel.get_audio_slicecCs|jS)zdGet sample rate. Returns: `int`: Sample rate of audio. )rr/rrrget_sample_rateszMel.get_sample_ratecCsntjj|||j|j|j|jd}tj|t j |j d}||j d|j  ddd t j}t|}|S)aConvert slice of audio to spectrogram. Args: slice (`int`): Slice number of audio to convert (out of `get_number_of_slices()`). Returns: `PIL Image`: A grayscale image of `x_res x y_res`. )yrrrr!)refrrg?)r&featuremelspectrogramr2rrrr! power_to_dbr)maxrclipastypeuint8r fromarray)rr1Slog_Sbytedataimagerrraudio_slice_to_images ( zMel.audio_slice_to_imagerBcCshtj|dd|j|jf}|d|jd|j}t |}tj j j ||j |j|j|jd}|S)zConverts spectrogram to audio. Args: image (`PIL Image`): An grayscale image of `x_res x y_res`. Returns: audio (`np.ndarray`): The audio as a NumPy array. r=)dtypefloatr6)rrrr)r) frombuffertobytesreshapeheightwidthr<rr& db_to_powerr7inverse mel_to_audiorrrr)rrBrAr@r?rrrrimage_to_audios  zMel.image_to_audio)r r r r r r r)NN)r)__name__ __module__ __qualname____doc__ config_namerintrrstrr)ndarrayr-r0r2r3rrCrNrrrrr%sB    r)numpyr)configuration_utilsrrschedulers.scheduling_utilsrr&rr ExceptionePILrrrrrrs