o
    5ôjD  ã                   @   s2  d dl Z d dlmZ d dlmZmZ d dlmZmZ d dl	Z
d dlZd dlm  mZ ddlmZ dZdZd	Zd
Zee ZeeeƒZed ZeeeƒZeeeƒZefdedefdd„Zefddœdedefdd„Zedd�dedejfdd„ƒZ 		 	d!deee
j!ejf dededeeeej"f  fdd „Z#dS )"é    N)Ú	lru_cache)ÚCalledProcessErrorÚrun)ÚOptionalÚUnioné   )Ú	exact_divi€>  i�  é    é   é   ÚfileÚsrc                 C   sˆ   ddddd| dddd	d
ddt |ƒdg}z
t|ddd�j}W n ty4 } ztd|j ¡ › �ƒ|‚d}~ww t |tj	¡ 
¡  tj¡d S )a?  
    Open an audio file and read as mono waveform, resampling as necessary

    Parameters
    ----------
    file: str
        The audio file to open

    sr: int
        The sample rate to resample the audio if necessary

    Returns
    -------
    A NumPy array containing the audio waveform, in float32 dtype.
    Zffmpegz-nostdinz-threadsÚ0z-iz-fZs16lez-acÚ1z-acodecZ	pcm_s16lez-arú-T)Úcapture_outputÚcheckzFailed to load audio: Ng      à@)Ústrr   Ústdoutr   ÚRuntimeErrorÚstderrÚdecodeÚnpÚ
frombufferÚint16ÚflattenÚastypeÚfloat32)r   r   ÚcmdÚoutÚe© r!   ú`/var/www/html/academy/tools/voices/clone/cosy-venv/lib/python3.10/site-packages/whisper/audio.pyÚ
load_audio   s"   ÷€ÿr#   éÿÿÿÿ)ÚaxisÚlengthr%   c                C   sæ   t  | ¡rC| j| |kr| j|t j|| jd�d�} | j| |k rAdg| j }d|| j|  f||< t | dd„ |ddd… D ƒ¡} | S | j| |krS| j	t
|ƒ|d	�} | j| |k rqdg| j }d|| j|  f||< t | |¡} | S )
zO
    Pad or trim the audio array to N_SAMPLES, as expected by the encoder.
    )Údevice)ÚdimÚindex)r   r   r   c                 S   s   g | ]	}|D ]}|‘qqS r!   r!   )Ú.0ÚsizesÚpadr!   r!   r"   Ú
<listcomp>N   s    zpad_or_trim.<locals>.<listcomp>Nr$   )Úindicesr%   )ÚtorchÚ	is_tensorÚshapeÚindex_selectÚaranger'   ÚndimÚFr,   ÚtakeÚranger   )Úarrayr&   r%   Z
pad_widthsr!   r!   r"   Úpad_or_trimA   s"   
ÿ 
ør9   )ÚmaxsizeÚn_melsÚreturnc                 C   sz   |dv sJ d|› �ƒ‚t j t j t¡dd¡}tj|dd��}t |d|› � ¡ 	| ¡W  d  ƒ S 1 s6w   Y  dS )	ad  
    load the mel filterbank matrix for projecting STFT into a Mel spectrogram.
    Allows decoupling librosa dependency; saved using:

        np.savez_compressed(
            "mel_filters.npz",
            mel_80=librosa.filters.mel(sr=16000, n_fft=400, n_mels=80),
            mel_128=librosa.filters.mel(sr=16000, n_fft=400, n_mels=128),
        )
    >   éP   é€   zUnsupported n_mels: Zassetszmel_filters.npzF)Úallow_pickleZmel_N)
ÚosÚpathÚjoinÚdirnameÚ__file__r   Úloadr/   Ú
from_numpyÚto)r'   r;   Zfilters_pathÚfr!   r!   r"   Úmel_filters[   s
   $ÿrI   r=   ÚaudioÚpaddingr'   c           
      C   sØ   t  | ¡st| tƒrt| ƒ} t  | ¡} |dur|  |¡} |dkr(t | d|f¡} t  	t
¡ | j¡}t j| t
t|dd�}|ddd…f  ¡ d }t| j|ƒ}|| }t j|dd	� ¡ }	t  |	|	 ¡ d
 ¡}	|	d d }	|	S )ap  
    Compute the log-Mel spectrogram of

    Parameters
    ----------
    audio: Union[str, np.ndarray, torch.Tensor], shape = (*)
        The path to audio or either a NumPy array or Tensor containing the audio waveform in 16 kHz

    n_mels: int
        The number of Mel-frequency filters, only 80 is supported

    padding: int
        Number of zero samples to pad to the right

    device: Optional[Union[str, torch.device]]
        If given, the audio tensor is moved to this device before STFT

    Returns
    -------
    torch.Tensor, shape = (80, n_frames)
        A Tensor that contains the Mel spectrogram
    Nr   T)ÚwindowÚreturn_complex.r$   r   g»½×Ùß|Û=)Úming       @g      @)r/   r0   Ú
isinstancer   r#   rF   rG   r5   r,   Úhann_windowÚN_FFTr'   ÚstftÚ
HOP_LENGTHÚabsrI   ÚclampÚlog10ÚmaximumÚmax)
rJ   r;   rK   r'   rL   rR   Z
magnitudesÚfiltersZmel_specZlog_specr!   r!   r"   Úlog_mel_spectrogramn   s"   



rZ   )r=   r   N)$r@   Ú	functoolsr   Ú
subprocessr   r   Útypingr   r   Únumpyr   r/   Ztorch.nn.functionalÚnnÚ
functionalr5   Úutilsr   ZSAMPLE_RATErQ   rS   ZCHUNK_LENGTHZ	N_SAMPLESZN_FRAMESZN_SAMPLES_PER_TOKENZFRAMES_PER_SECONDZTOKENS_PER_SECONDr   Úintr#   r9   ÚTensorrI   Úndarrayr'   rZ   r!   r!   r!   r"   Ú<module>   sB    


(üÿþýü