m\iWSrSSKrSSKrSSKrSSKrSSKrSSKrSSKJr SSK J r J r J r J r SSKrSSKrSSKrSSKJr SS\ \R(\R*4S\S\S \ \R(\R*44S jjrSS\ \R(\R*4S \S \ \R(\R*44S jjr"S S5rS \4SjrS \4SjrS \4SjrSS\ \R(\R*4S\ \S \4Sjjr\"SS9r S S\ \R(\R*4S\ \\4S\S\ \S\!S\ \S\ \S \4Sjjr"g)!z Audio saving and transcoding utility module Independent audio file operations outside of handler, supporting: - Save audio tensor/numpy to files (default FLAC format, fast) - Format conversion (FLAC/WAV/MP3) - Batch processing N)Path)UnionOptionalListTuple)logger audio_datafade_in_samplesfade_out_samplesreturnc0US::aUS::aU$[U[R5nU(a UR5nURSnOUR 5nURSnUS:a[ X5nU(a?[R"SSXdRURS9nUSSU24U-USSU24'O8[R"SSU[RS9nUSSU24U-USSU24'US:a[ X%5nU(aD[R"SSXRURS9nUSXX- S24U-USXX- S24'U$[R"SSU[RS9nUSXX- S24U-USXX- S24'U$) amApply linear fade in and/or fade out to audio data. Args: audio_data: Audio data as torch.Tensor [channels, samples] or numpy.ndarray. fade_in_samples: Number of samples for fade in ramp (0 = no fade in). fade_out_samples: Number of samples for fade out ramp (0 = no fade out). Returns: Audio data with fades applied, in the same format as input. rgg?)dtypedevice.N)r) isinstancetorchTensorcloneshapecopyminlinspacerrnpfloat32) r r r is_tensoraudio total_samples actual_inramp actual_outs 2/mnt/workspace/ACE-Step-1.5/acestep/audio_utils.py apply_fader"s! 0A 5:u||4I  " B ! B 7 >>#sI[[QVQ]Q]^D%*3  ?%;d%BE#z z/ ";;sC"**ED%*3  ?%;d%BE#z z/ "!)9 >>#sJkkRWR^R^_D6;CA[A\<\6]`d6dE#}122 3 L;;sC2::FD6;CA[A\<\6]`d6dE#}122 3 L target_dbcv[U[R5(aUR5nSnOUR 5nSnU(a+[R "[R "U55nO*[R "[R "U55nUS:aU$SUS- -nXT- nX&-nU$)z Apply peak normalization to audio data. Args: audio_data: Audio data as torch.Tensor or numpy.ndarray target_db: Target peak level in dB (default: -1.0) Returns: Normalized audio data in the same format as input TFgư> g4@)rrrrrmaxabsr)r r$rrpeak target_ampgains r!normalize_audior,Hs*ell++  " ! yy5)*vvbffUm$ d{ D()J  D LE Lr#c\rSrSrSrSr1SkrSrSS1rSS\ 4Sjjr S S \ RS \ S \S \\ S\\SS 4 SjjrS!S\\ R\R&4S \\ \ 4S\S\\ S\S \\ S\\S\ 4SjjrS"S\\ \ 4S \\ \ 4S\ S\S \\ S\\S\ 4SjjrS#S\\\ R\ R4S\\ \ 4S\ S\S\\ S\S \\ S\\S\\ 4SjjrSrg )$ AudioSaverrz*Audio saving and transcoding utility class128k>r0192k256k320k逻iDdefault_formatcUR5UlURS;a"[R"SUS35 SUlgg)zv Initialize audio saver Args: default_format: Default save format ('flac', 'wav', 'mp3', 'wav32', 'opus', 'aac') flacwavmp3wav32opusaacUnsupported format z, using 'flac'r8N)lowerr5rwarning)selfr5s r!__init__AudioSaver.__init__zsI-224   &T T NN00@O P"(D  Ur#N audio_tensor output_pathinput_sample_rate mp3_bitratemp3_sample_rater c[U=(d UR5R5R5nX`R;a URn[ U=(d UR 5nXpR;a UR nUn[ U5[ U5:wa3[RRU[ U5[ U55n[R"SSS9n [U R5n SSS5 [R "[W 5U[ U5SSS9 SS S S S S [U 5SSS[[ U55SU[U5/n ["R$"U SSSS9 [&R("SUSUSUS35 U R7SS9 g![a UR nGN\f=f!,(df  N=f![*an [-S5U eSn A f["R.an [-S5U eSn A f["R0aJn U R2(aU R2R5SSS9O [U 5n [-SU 35U eSn A ff=f![a [&R8"SU 35 gf=f!W R7SS9 f![a [&R8"SW 35 ff=f=f) z=Save MP3 with explicit ffmpeg settings and 128k/48k defaults..wavF)suffixdeleteNT soundfilechannels_firstbackendffmpegz-yz -hide_bannerz -loglevelerrorz-iz-codec:a libmp3lamez-arz-b:ax)checkcapture_outputtimeout[AudioSaver] Saved audio to z (mp3, zHz, )zRffmpeg executable not found. Install ffmpeg or add it to PATH to export MP3 files.z.ffmpeg MP3 export timed out after 120 seconds.utf-8ignore)errorszffmpeg MP3 export failed: ) missing_okz2[AudioSaver] Failed to remove temporary WAV file: )strMP3_DEFAULT_BITRATEstripr?MP3_ALLOWED_BITRATESintMP3_DEFAULT_SAMPLE_RATE ExceptionMP3_ALLOWED_SAMPLE_RATES torchaudio functionalresampletempfileNamedTemporaryFilernamesave subprocessrunrdebugFileNotFoundError RuntimeErrorTimeoutExpiredCalledProcessErrorstderrdecodeunlinkr@)rArDrErFrGrHbitratetarget_sample_ratetensor_to_savetemp_wav temp_wav_pathcmderts r! _save_mp3AudioSaver._save_mp3sk=T%=%=>DDFLLN 33 3..G >!$_%T8T8T!U  %B%B B!%!=!= %  !S);%< <'22;;L#N_J`befxbyzN  ( (u E /MF e OOM"&'##  $ Wc-(Ls3123K C NN3d4 M LL7 }GL^K__cdkcllmn o e$$$5O >!%!=!=  >F E*! |stz{ {(( XOPVW W,, MBC((QXX__WX_>PSTUPVF!;F8DE1 L M  e!STaSbcd e e$$$5 e!STaSbcd esF82GBG&(J8GG G#& J0 G<<J HJ6AI;;JJ,#J)(J),K'.J>=K'>#K$!K'#K$$K'r sample_rateformatrOc U=(d URR5nUS;a2[R"SUSUR35 URn[ U5nUS:XaSOSU3nUR R5S;aUR U5nO[US:Xa0UR R5S:XaUR S5nO%US :XaUR R5S :Xa[U[R5(aU(a%[R"U5R5n O[R"U5R5n U R5S :Xa,U RS U RS :a U Rn OeUR!5R5n U(d@U R5S :Xa,U RS U RS :a U Rn U R#5n US:XaUR%U UUUUS9 ['U5$US;a"[(R*"['U5U USSS9 OUS;aUS:XahS SKn U R/S S 5R15n U R3['U5XSSS9 [R4"SUSUS35 ['U5$[(R*"['U5U USSS9 O [(R*"['U5U USS 9 [R4"SUS!US"US35 ['U5$![6a&n [R8"SU S35 SnSn A NSn A ff=f![6an US:Xa[R8"S#U 35 eS SKn U R/S S 5R15n US:XaSn SnOUR;5n SnU R3['U5XXS$9 [R4"S%US!US"US35 ['U5sSn A $![6a"n[R8"S&U S'U35 eSnAff=fSn A ff=f)(a7 Save audio data to file Args: audio_data: Audio data, torch.Tensor [channels, samples] or numpy.ndarray output_path: Output file path (extension can be omitted) sample_rate: Sample rate format: Audio format ('flac', 'wav', 'mp3', 'wav32', 'opus', 'aac'), defaults to default_format channels_first: If True, tensor format is [channels, samples], else [samples, channels] mp3_bitrate: Optional MP3 bitrate override (128k/192k/256k/320k) mp3_sample_rate: Optional MP3 sample rate override (44100/48000) Returns: Actual saved file path r7r>z, using r;rJ.)z.flacrJz.mp3z.opusz.aac.m4az.wav32r=rrr:)rGrH)r<r=TrQrN)r8r9r;NFLOATWAV)subtyperrXz (wav32, zHz)zFailed to save wav32: z, falling back to standard wavr9rM)rOz (z, z1[AudioSaver] MP3 export failed without fallback: )rrz/[AudioSaver] Fallback soundfile Saved audio to z#[AudioSaver] Failed to save audio: z -> Fallback failed: )r5r?rr@rrK with_suffixrrndarrayr from_numpyfloatdimrTcpu contiguousr~r^rfrlrM transposenumpywriterordrRupper)rAr rErrrOrGrHextrDsfaudio_npr} sf_formatrinner_es r! save_audioAudioSaver.save_audiosE2/D//668 G G NN0ATAT@UV W((F;' ')f6(|    # # %-_ _%11#6K w ;#5#5#;#;#=#I&226:[ u_!3!3!9!9!;v!E  j"** - -$// ;AAC  %// ;AAC ##%*|/A/A!/D|GYGYZ[G\/\$0NN\&>>+113L!l&6&6&8A&=%%a(<+=+=a+@@#/>>L$..0 N   +$3 ;''?* $ #'$ 33W$ '.$0#9#9!Q#?#E#E#G[!18RYbgh 'CK=PYZeYffi%jk";//  $ #''  $ #'  LL7 }BvhbQ\P]]`a b{# #-%' '=aS@^%_`!&'0  PQRPSTU &'11!Q7==?W$ %I%G & I"G[)8d N{m[]^d]eeghsgttwxy;''  B1#EZ[bZcde ' so%#N 4N>A&M%A,N NM=8N=NN Q/ Q*0BP;5Q/; Q'Q""Q''Q**Q/ input_path output_format remove_inputc D[U5n[U5nUR5(d[SU35e[R"[ U55upxUR UUUUSUUS9nU(a)UR5 [R"SU35 U$)a Convert audio format Args: input_path: Input audio file path output_path: Output audio file path output_format: Target format ('flac', 'wav', 'mp3', 'wav32', 'opus', 'aac') remove_input: Whether to delete input file mp3_bitrate: Optional MP3 bitrate override (128k/192k/256k/320k) mp3_sample_rate: Optional MP3 sample rate override (44100/48000) Returns: Output file path zInput file not found: TrrrOrGrHz![AudioSaver] Removed input file: ) rexistsrprfloadr^rrvrro) rArrErrrGrHrDrs r! convert_audioAudioSaver.convert_audioTs.*% ;'   ""#&L<#8#8~~HI|4499;< $$\2HHV$   77>>) $ $   i ( WW]]9 % %  Y677>>%((%%e, E %&! !w$! . ' !  )=@$-8   4"4'&  s*1 G(=.G+ G( G% !G(( G65G6cTUcg[U[5(a[RR U5(aH[ US5n[ R"UR55R5sSSS5 $[ R"URS55R5$[US5(aF[ R"[UR5RS55R5$[ R"[U5RS55R5$!,(df  N=f![a? [ R"[U5RS55R5s$f=f)z Get hash identifier for an audio file. Args: audio_file: Path to audio file (str) or file-like object Returns: Hash string or empty string NrrrZrk)rr^rrrrrrrrencodehasattrrkrd) audio_filers r!get_audio_file_hashrs' K j# & &ww~~j))*d+q">>!&&(3==?,+>>*"3"3G"<=GGI I Z ( (>>#joo"6"="=g"FGQQS S~~c*o44W=>HHJJ ,+ K~~c*o44W=>HHJJKs>AE 2E = E2E:AE;E EEAF'&F'c [R"USSS9n[R"UR S55nUR 5nUSSSUSSSUSS SUS S SUS S 3 nU$) z Generate deterministic UUID from generation parameters. Same parameters will always generate the same UUID. Args: params_dict: Dictionary of parameters Returns: UUID string TF) sort_keys ensure_asciirZr-  )jsondumpsrrrr) params_dict params_jsonrhash_hexuuid_strs r!generate_uuid_from_paramsrs**[DuMK~~k009:H!!#H1Q-(1R.!18Br?2C1XbQS_DUUVW_`bceWfVghH Or#seedcb[U[R5(aUR5R 5nOUn[ R "UR55R5nUb9USU3n[ R "UR55R5$U$)z Generate UUID from audio data (for caching/deduplication) Args: audio_data: Audio data seed: Optional seed value Returns: UUID string r) rrrrrrrtobytesrr)r rr data_hashcombineds r!generate_uuid_from_audio_datar's*ell++>>#))+x//12<<>I [$(~~hoo/0::<< r#r8rrErrrOrGrHc 8[RUUUUUUU5$)a Convenience function: save audio (using default configuration) Args: audio_data: Audio data output_path: Output path sample_rate: Sample rate format: Format (default flac) channels_first: Tensor format flag mp3_bitrate: Optional MP3 bitrate override (128k/192k/256k/320k) mp3_sample_rate: Optional MP3 sample rate override (44100/48000) Returns: Saved file path )_default_saverr)r rErrrOrGrHs r!rrIs,0  $ $ r#)rr)g)Nr)#riorrrmrripathlibrtypingrrrrrrrrflogururrrrbr"rr,r.r^rrrrrrrrr#r!r s //  ,ellBJJ./,,, 5<< #$ ,^&ellBJJ&> ?&E&]bchcocoqsq{q{c{]|&TIIX 95#95xKsK4c*ellBJJ./ 3- >62  !%%) ellBJJ./ sDy!   SM    # c]   r#