biddlZddlmZmZddlmZ ddlZdZdZ dd l m Z Gd d eeZy#e $rZ dZde d Z YdZ [ %dZ [ wwxYw) N) ConfigMixinregister_to_config)SchedulerMixinTFzCannot import librosa because zB. Make sure to correctly install librosa to be able to install it.)Imagec"eZdZdZdZe ddedededededed efd Zdedefd Zdd e de jfdZ defdZ ddede jfdZdefdZdedej fdZdej de jfdZy )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.jsonx_resy_res sample_raten_fft hop_lengthtop_dbn_iterc||_||_||_||_||_|j ||d|_tstty)N) rsrrrrset_resolutionaudio_librosa_can_be_imported ValueError _import_error)selfr r r rrrrs m/home/cdr/jupyterlab/.venv/lib/python3.12/site-packages/diffusers/pipelines/deprecated/audio_diffusion/mel.py__init__z Mel.__init__:sQ%    E5) ']+ +(c||_||_|j|_|j|jzdz |_y)zSet resolution. Args: x_res (`int`): x resolution of spectrogram (time). y_res (`int`): y resolution of spectrogram (frequency bins). N)r r n_melsr slice_size)rr r s rrzMel.set_resolutionPs6  jj **t6:rN audio_file raw_audioc |+tj|d|j\|_}n||_t |j|j |j zkrhtj|jtj|j |j zt |jz fg|_yy)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) librosaloadrrlenr rnp concatenatezeros)rr!r"_s r load_audiozMel.load_audio^s  !#LL$477KMDJ"DJ tzz?TZZ$//9 9RXXtzzDOO?[^abfblbl^m?m>o5p(qrDJ :rreturncFt|j|jzS)zGet number of slices in audio. Returns: `int`: Number of spectograms audio can be sliced into. )r'rr rs rget_number_of_sliceszMel.get_number_of_slicesps4::$//11rslicecZ|j|j|z|j|dzzS)zGet slice of audio. Args: slice (`int`): Slice number of audio (out of `get_number_of_slices()`). Returns: `np.ndarray`: The audio slice as a NumPy array. r)rr )rr1s rget_audio_slicezMel.get_audio_sliceys+zz$//E1DOOuqy4QRRrc|jS)zdGet sample rate. Returns: `int`: Sample rate of audio. )rr/s rget_sample_ratezMel.get_sample_rates wwrctjj|j||j|j |j |j}tj|tj|j}||jzdz|jz jdddzjtj}tj |}|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%featuremelspectrogramr3rrrr power_to_dbr(maxrclipastypeuint8r fromarray)rr1Slog_Sbytedataimages raudio_slice_to_imagezMel.audio_slice_to_images OO * *""5)dggTZZTXTcTclplwlw + ##A266$++Fdkk)S04;;>DDQLsRZZ[][c[cd) rrEctj|jdj|j|j f}|j d|jzdz |jz }tj|}tjjj||j|j|j|j }|S)zConverts spectrogram to audio. Args: image (`PIL Image`): An grayscale image of `x_res x y_res`. Returns: audio (`np.ndarray`): The audio as a NumPy array. r@)dtypefloatr9)rrrr)r( frombuffertobytesreshapeheightwidthr?rr% db_to_powerr:inverse mel_to_audiorrrr)rrErDrCrBrs rimage_to_audiozMel.image_to_audios==@HH%,,X]XcXcIde(4;;6resb C: #MN+~N$ (+mns/AAA