j> ddlZddlZddlmZddlmZmZddlZddl Z ddl m cm Z ddlmZdZdZdZdZeezZeeeZed zZeeeZeeeZefd ed ed ejfd ZefdddedefdZedded e jfdZ ddeeeje jfdededeeee j ffdZ!dS)N) lru_cache)OptionalUnion) exact_divi>ifilesrreturnc ddddd|dddd d d d t|d g}tj|ddj}nA#tj$r/}t d|j|d}~wwxYwtj |tj  tj dz S)a? Open an audio file and read as mono waveform, resampling as necessary Parameters ---------- file: str The audio file to open sr: int The sample rate to resample the audio if necessary Returns ------- A NumPy array containing the audio waveform, in float32 dtype. ffmpegz-nostdinz-threads0z-iz-fs16lez-ac1z-acodec pcm_s16lez-ar-T)capture_outputcheckzFailed to load audio: Ng@)str subprocessrunstdoutCalledProcessError RuntimeErrorstderrdecodenp frombufferint16flattenastypefloat32)r r cmdoutes q/home/lighthouse/badge-speaker-benchmark/.benchmark-venv-community/lib/python3.11/site-packages/whisperx/audio.py load_audior(s P              GG  "nSTBBBI  (PPPGAHOO4E4EGGHHaOP =bh ' ' / / 1 1 8 8 D Dw NNs:=A; *A66A;)axislengthr*cZtj|r|j||kr0||tj||j}|j||krHdg|jz}d||j|z f||<tj|d|dddD}n{|j||kr$| t||}|j||kr5dg|jz}d||j|z f||<tj||}|S) zO Pad or trim the audio array to N_SAMPLES, as expected by the encoder. )device)dimindex)rrrcg|] }|D]}| Sr1).0sizespads r' zpad_or_trim..Qs%!U!U!U%u!U!U#!U!U!U!UNr))indicesr*) torch is_tensorshape index_selectaranger-ndimFr4takeranger)arrayr+r* pad_widthss r' pad_or_trimrCDs? u. ;t v % %&& VEL I I I'E ;t v % % EJ.J !6EK,=#=>Jt E%!U!U:ddd3C!U!U!UVVE ;t v % %JJuV}}4J@@E ;t v % % EJ.J !6EK,=#=>Jt F5*--E Lr6)maxsizen_melscb|dvs Jd|tjtjtjt dd5}tj|d| |cdddS#1swxYwYdS)a 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), ) )PzUnsupported n_mels: assetszmel_filters.npzmel_N) rloadospathjoindirname__file__r8 from_numpyto)r-rEfs r' mel_filtersrT^s Y    ?v ? ?      RW__X..:KLL  ? /// 23366v>>??????????????????s'0B$$B(+B(audiopaddingr-ctj|s8t|trt |}tj|}|||}|dkrtj|d|f}tj t|j }tj |tt|d}|dddfdz}t|j |}||z}tj|d } tj| | d z } | d zd z } | S) ap Compute the log-Mel spectrogram of Parameters ---------- audio: Union[str, np.ndarray, torch.Tensor], shape = (*) The path to audio or either a NumPy array or Tensor containing the audio waveform in 16 kHz n_mels: int The number of Mel-frequency filters, only 80 is supported padding: int Number of zero samples to pad to the right device: Optional[Union[str, torch.device]] If given, the audio tensor is moved to this device before STFT Returns ------- torch.Tensor, shape = (80, n_frames) A Tensor that contains the Mel spectrogram NrT)windowreturn_complex.r)r g|=)ming @g@)r8r9 isinstancerr(rQrRr>r4 hann_windowN_FFTr-stft HOP_LENGTHabsrTclamplog10maximummax) rUrErVr-rXr^ magnitudesfiltersmel_speclog_specs r'log_mel_spectrogramripsB8 ?5 ! !( eS ! ! &u%%E ''   {{ea\**  u % % ( ( 6 6F :eUJvd S S SDc3B3h##%%*J%,//G#H{8///5577H}Xx||~~';<whisperx.utilsr SAMPLE_RATEr]r_ CHUNK_LENGTH N_SAMPLESN_FRAMESN_SAMPLES_PER_TOKENFRAMES_PER_SECONDTOKENS_PER_SECONDrintndarrayr(rCTensorrTr-rir1r6r'r{s  """""""" $$$$$$     ; & 9Y + + 1nIk:66Ik+>??%0(O(OS(Oc(OBJ(O(O(O(OV&/rs4 4?? ????(15 // bj%,. // // U3 ,- . //////r6