shQ SSKrSSKJr SSKJrJr SSKJrJr SSK r SSK r SSK J s Jr SSKJr SrSrS rS r\\-r\"\\5r\S -r\"\\5r\"\\5r\4S \S \4Sjjr\4SS.S\S\4Sjjjr\"SS9S\S\ R>4Sj5r SS\\\ RB\ R>4S\S\S\\\\ RD44Sjjr#g)N) lru_cache)CalledProcessErrorrun)OptionalUnion) exact_divi>ifilesrczSSSSSUSSSS S S S [U5S /n[USSS9Rn[R"U[R5R5R[R5S- $![a,n[ SUR R 535UeSnAff=f)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@) strrstdoutr RuntimeErrorstderrdecodenp frombufferint16flattenastypefloat32)r rcmdoutes a/Users/tiagomarins/Projetos/claudeai/copy_bank/venv/lib/python3.13/site-packages/whisper/audio.py load_audior's* C d g s; s2w  CP#d$7>> ==bhh ' / / 1 8 8 Dw NN P3AHHOO4E3FGHaOPsB B:'B55B:)axislengthr)c [R"U5(aURUU:a,URU[R"XR S9S9nURUU:a\S/UR -nSXRU- 4X2'[R"XSSS2VVs/sH oDHoUPM M snn5nU$URUU:aUR[U5US9nURUU:a;S/UR -nSXRU- 4X2'[R"X5nU$s snnf)zG Pad or trim the audio array to N_SAMPLES, as expected by the encoder. )device)dimindex)rrrNr()indicesr)) torch is_tensorshape index_selectaranger,ndimFpadtakeranger)arrayr*r) pad_widthssizesr7s r& pad_or_trimr=As: u ;;t v %&& VLL I'E ;;t v % EJJ.J !6KK,=#=>J EE%dd3C!U3C%u#u#3C!UVE L ;;t v %JJuV}4J@E ;;t v % EJJ.J !6KK,=#=>J FF5-E L"Vs.E )maxsizen_melsreturnc\US;d SU35e[RR[RR[5SS5n[ R "USS9n[R"USU35RU5sSSS5 $!,(df  g=f) aD load the mel filterbank matrix for projecting STFT into a Mel spectrogram. Allows decoupling librosa dependency; saved using: np.savez_compressed( "mel_filters.npz", mel_80=librosa.filters.mel(sr=16000, n_fft=400, n_mels=80), mel_128=librosa.filters.mel(sr=16000, n_fft=400, n_mels=128), ) >PzUnsupported n_mels: assetszmel_filters.npzF) allow_picklemel_N) ospathjoindirname__file__rloadr0 from_numpyto)r,r? filters_pathfs r& mel_filtersrQ[s Y ?"6vh ?? 77<< 98EVWL E 2aD/ 2366v> 3 2 2s (+B B+audiopaddingr,c[R"U5(d6[U[5(a [ U5n[R "U5nUbUR U5nUS:a[R"USU45n[R"[5R UR5n[R"U[[USS9nUSSS24R5S-n[URU5nXv-n[R "USS 9R#5n [R$"XR'5S - 5n U S -S - n U $) a= 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 and 128 are 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 = (n_mels, n_frames) A Tensor that contains the Mel spectrogram NrT)windowreturn_complex.r(r g|=)ming @g@)r0r1 isinstancerr'rMrNr6r7 hann_windowN_FFTr,stft HOP_LENGTHabsrQclamplog10maximummax) rRr?rSr,rUr[ magnitudesfiltersmel_speclog_specs r&log_mel_spectrogramrfns8 ??5 ! ! eS ! !u%E  '  {ea\*   u % ( ( 6F ::eUJvd SDc3B3h##%*J%,,/G#H{{8/557H}}X||~';r{s2 ."      ; & Y + 1nk:6k+>?%0%OS%Oc%OP&/rs4 4?? ??(15 / bjj%,,. // // U3 ,- . /rg