i dZddlZddlZddlZddlZddlZddlZddlZddlZddl Z ddl Z ddl Z ddl Z ddl Z ddlmZddlmZmZmZmZddlmZddlmZejeZddlmZddlmZm Z m!Z!dd l"m#Z#d Z$d Z%d Z&d Z'dZ(dZ)dZ*dZ+dZ,dZ-dZ.dZ/dZ0dZ1dZ2dZ3dZ4dZ5dZ6dZ7dZ8dZ9d Z:d!Z;d"Zd$Z?d%Z@d&eAfd'ZBeBZCd(ZDd&eeAeffd)ZEd*eeAefd&eAfd+ZFd&eGfd,ZHd-eAd&eeAfd.ZId/eAd0eAd*eeAefd&eAfd1ZJd/eAd0eAd*eeAefd&eAfd2ZKd/eAd0eAd*eeAefd&eAfd3ZLd/eAd0eAd*eeAefd&eAfd4ZMd/eAd0eAd*eeAefd&eAfd5ZNd/eAd0eAd*eeAefd&eAfd6ZOe>e?e@fd7ePd8eQd9eQd:eQd&ePf d;ZRd/eAd0eAd*eeAefd&eAfd<ZSd&eGfd=ZTd&eAfd>ZUd&eAfd?ZVd/eAd0eAd*eeAefd&eAfd@ZW dd/eAd0eeAd&eAfdAZXd&eGfdBZYd&eZeAeAffdCZ[d&eGfdDZ\ej]dEZ^ej]dFZ_ej]dGZ`ej]dHZaej]dIZbej]dJZcej]dKZdej]dLejeMZfej]dNejeMZgej]dOZhej]dPZid/eAd&eAfdQZj ddRejkdSe jldTe jldUeeeAgdffdVZmedWkrendXendYdZZoend[end\eoe$drd]nd^end_eoe%d`rd]ndaendbejpdcrddndeendfeoe&dgrd]ndhendbe!rddndiendjejpdkrdlndmendneHrdondpendqeCeEZqeFeqZrendrerddslsmtZtmuZudtdudvdwdxdydwdzed{dyd|d/gd}d~ZvetjwdtdevdeYddS)a Text-to-Speech Tool Module Supports seven TTS providers: - Edge TTS (default, free, no API key): Microsoft Edge neural voices - ElevenLabs (premium): High-quality voices, needs ELEVENLABS_API_KEY - OpenAI TTS: Good quality, needs OPENAI_API_KEY - MiniMax TTS: High-quality with voice cloning, needs MINIMAX_API_KEY - Mistral (Voxtral TTS): Multilingual, native Opus, needs MISTRAL_API_KEY - Google Gemini TTS: Controllable, 30 prebuilt voices, needs GEMINI_API_KEY - NeuTTS (local, free, no API key): On-device TTS via neutts_cli, needs neutts installed Output formats: - Opus (.ogg) for Telegram voice bubbles (requires ffmpeg for Edge TTS) - MP3 (.mp3) for everything else (CLI, Discord, WhatsApp) Configuration is loaded from ~/.hermes/config.yaml under the 'tts:' key. The user chooses the provider and voice; the model just sends text. Usage: from tools.tts_tool import text_to_speech_tool, check_tts_requirements result = text_to_speech_tool(text="Hello world") N)Path)CallableDictAnyOptional)urljoin)display_hermes_home)resolve_managed_tool_gateway)managed_nous_tools_enabledprefers_gatewayresolve_openai_audio_api_key)hermes_xai_user_agentcddl}|S)z?Lazy import edge_tts. Returns the module or raises ImportError.rNedge_ttsrs 6/home/agentuser/.hermes/hermes-agent/tools/tts_tool.py_import_edge_ttsr8sOOO Ocddlm}|S)zGLazy import ElevenLabs client. Returns the class or raises ImportError.r ElevenLabs)elevenlabs.clientrrs r_import_elevenlabsr=s,,,,,, rcddlm}|S)zCLazy import OpenAI client. Returns the class or raises ImportError.r)OpenAI)openair) OpenAIClients r_import_openai_clientrBs------ rcddlm}|S)zDLazy import Mistral client. Returns the class or raises ImportError.rMistral)mistralai.clientr!r s r_import_mistral_clientr#Gs(((((( Nrcddl}|S)zJLazy import sounddevice. Returns the module or raises ImportError/OSError.rN) sounddevice)sds r_import_sounddevicer'Ls Iredgezen-US-AriaNeuralpNInz6obpgDQGcFmaJgBeleven_multilingual_v2eleven_flash_v2_5zgpt-4o-mini-ttsalloyzhttps://api.openai.com/v1z speech-2.8-hdEnglish_Graceful_Ladyz https://api.minimax.io/v1/t2a_v2zvoxtral-mini-tts-2603z$c69964a6-ab8b-4f8a-9465-ec0925096ec8eveen]zhttps://api.x.ai/v1zgemini-2.5-flash-preview-ttsKorez0https://generativelanguage.googleapis.com/v1betareturnc@ddlm}t|ddS)Nrget_hermes_dirz cache/audio audio_cache)hermes_constantsr8strr7s r_get_default_output_dirr<os.////// ~~m];; < <get ImportErrorloggerdebug Exceptionwarning)r>configes r_load_tts_configrKzs 111111zz%$$$  PQQQ 6DIII s!%(&B BA=7B=B tts_configc|dptS)z%Get the configured TTS provider name.provider)rCDEFAULT_PROVIDERlowerstrip)rLs r _get_providerrRs2 NN: & & :*: A A C C I I K KKrc.tjdduS)z+Check if ffmpeg is available on the system.ffmpegN)shutilwhichrr _has_ffmpegrXs < ! ! --rmp3_pathctsdS|ddddz} tjdd|dd d d d d dd|dg dd}|jdkrEt d|j|jdddddStj |r%tj |dkr|Sn#tj $rt dYnXt$rt dYn3t$r'}t d|dYd}~nd}~wwxYwdS)z Convert an MP3 file to OGG Opus format for Telegram voice bubbles. Args: mp3_path: Path to the input MP3 file. Returns: Path to the .ogg file, or None if conversion fails. N.r3r.oggrT-i-acodeclibopus-ac1-b:a64k-vbroff-yTcapture_outputtimeoutz0ffmpeg conversion failed with return code %d: %sutf-8ignoreerrorsz)ffmpeg OGG conversion timed out after 30szffmpeg not found in PATHz ffmpeg OGG conversion failed: %sr@)rXrsplit subprocessrun returncoderErHstderrdecodeospathexistsgetsizeTimeoutExpiredFileNotFoundErrorrG)rYogg_pathresultrJs r_convert_to_opusr~s ==tsA&&q)F2HM tXy) Cx G      ! ! NNM +V]-A-A'RZ-A-[-[\`]`\`-a c c c4 7>>( # # (A(AA(E(EO  $DDDBCCCCC 333122222 MMM91tLLLLLLLLM 4s+A2C*%AC**)E*#E*: E*E%%E*text output_pathc Kt}|di}|dt}t|d|dd}d|i}|dkrt |dz dz}|dd|d<|j|fi|} | |d {V|S) z Generate audio using Edge TTS. Args: text: Text to convert. output_path: Where to save the MP3 file. tts_config: TTS config dict. Returns: Path to the saved audio file. r(voicespeed?dz+d%rateN)rrCDEFAULT_EDGE_VOICEfloatround Communicatesave) rrrL _edge_tts edge_configrrkwargspct communicates r_generate_edge_ttsrs!""I..,,K OOG%7 8 8E +//':>>'3+G+GHH I IEu F ||US[C'((v')'7777K   ; ' '''''''' rctjdd}|std|di}|dt}|dt }|drd}nd }t}|| } | j |||| } t|d 5} | D]} | |  d d d n #1swxYwY|S)z Generate audio using ElevenLabs. Args: text: Text to convert. output_path: Where to save the audio file. tts_config: TTS config dict. Returns: Path to the saved audio file. ELEVENLABS_API_KEYz=ELEVENLABS_API_KEY not set. Get one at https://elevenlabs.io/ elevenlabsvoice_idmodel_idr\ opus_48000_64 mp3_44100_128api_keyrrr output_formatwbN) rvgetenv ValueErrorrCDEFAULT_ELEVENLABS_VOICE_IDDEFAULT_ELEVENLABS_MODEL_IDendswithrtext_to_speechconvertopenwrite) rrrLr el_configrrrrclientaudio_generatorfchunks r_generate_elevenlabsrsdi,b11G ZXYYY|R00I}}Z)DEEH}}Z)DEEHF##(' ' #%%J Z ( ( (F+33 # 4O k4 A$  E GGENNNN  sC==DDc t\}}|di}|dt}|dt}|d|}t |d|dd}|drd} nd } t } | || } t|||| d ttj i } |dkr!td td|| d<| j jjdi| } | ||t#| dd}t%|r |SS#t#| dd}t%|r |wwxYw)z Generate audio using OpenAI TTS. Args: text: Text to convert. output_path: Where to save the audio file. tts_config: TTS config dict. Returns: Path to the saved audio file. rmodelrbase_urlrrr\opusmp3)rrzx-idempotency-key)rrinputresponse_format extra_headersg?g@closeNrW)#_resolve_openai_audio_client_configrCDEFAULT_OPENAI_MODELDEFAULT_OPENAI_VOICErrrdictr;uuiduuid4maxminaudiospeechcreatestream_to_filegetattrcallable)rrrLrr oai_configrrrrrr create_kwargsresponsers r_generate_openai_ttsr s<==GX"--J NN7$8 9 9E NN7$8 9 9E~~j(33H *..*..#*F*FGG H HEF##  (**L \'H = = =F+.DJLL0A0AB     C<<%(s3%?%?M' "-6<&->> >> ,,,.. E??  EGGGG .. E??  EGGGG s !BF-Gcddl}tjdd}|st d|di}t |dtpt}t |dtpt}t|d t}t|d t} t |d ptjd pt d } |drdnd} |||d} | dks|tks| dkr(| tkrd| i} |r|| d <| dkr| r| | d <| | d<|| dd|dt!d| d}|t%|d5}||jdddn #1swxYwY|S)z Generate audio using xAI TTS. xAI exposes a dedicated /v1/tts endpoint instead of the OpenAI audio.speech API shape, so this is implemented as a separate backend. rN XAI_API_KEYrz5XAI_API_KEY not set. Get one at https://console.x.ai/xairlanguage sample_ratebit_rater XAI_BASE_URL/.wavwavr)rrrcodecrz/ttsBearer application/json) Authorization Content-Typez User-Agent<)headersjsonrjr)requestsrvrrQrrCr;DEFAULT_XAI_VOICE_IDDEFAULT_XAI_LANGUAGEintDEFAULT_XAI_SAMPLE_RATEDEFAULT_XAI_BIT_RATEDEFAULT_XAI_BASE_URLrstriprpostrraise_for_statusrrcontent)rrrLrr xai_configrrrrrrpayloadrrrs r_generate_xai_ttsr?sOOOi r**0022G RPQQQr**J:>>*.BCCDDJJLLdPdH:>>*.BCCDDJJLLdPdHjnn]4KLLMMK:>>*.BCCDDHz"" 9^ $ $  eggffSkk  !))&11 >h>(0M* %#0 }} 0w00./11    H  k4 "A  !!!""""""""""""""" s8II#&I#c ddl}tjdd}|std|di}|dt }|dt }|d |d d }|d d } |d d} |d t} |drd} n|drd} nd} ||d||| | ddd| d dd} dd|d}| | | |d}| | }|di}|dd }|dkr+|d!d"}td#|d$||d%id&d}|std't|}t|d(5}||dddn #1swxYwY|S))a Generate audio using MiniMax TTS API. MiniMax returns hex-encoded audio data. Supports streaming (SSE) and non-streaming modes. This implementation uses non-streaming for simplicity. Args: text: Text to convert (max 10,000 characters). output_path: Where to save the audio file. tts_config: TTS config dict. Returns: Path to the saved audio file. rNMINIMAX_API_KEYrz@MINIMAX_API_KEY not set. Get one at https://platform.minimax.io/minimaxrrrr3volpitchrrr.flacflacrF)rrrri}r1)rbitrateformatchannel)rrstream voice_setting audio_settingrr)rrr)rrrj base_resp status_code status_msg unknown errorzMiniMax TTS API error (code ): datarz%MiniMax TTS returned empty audio datar)rrvrrrCDEFAULT_MINIMAX_MODELDEFAULT_MINIMAX_VOICE_IDDEFAULT_MINIMAX_BASE_URLrrrr RuntimeErrorbytesfromhexrr)rrrLrr mm_configrrrrrr audio_formatrrrr}rrr hex_audio audio_bytesrs r_generate_minimax_ttsrsOOOi)2..G ][\\\y"--I MM'#8 9 9E}}Z)ABBH MM':>>'1#=#= > >E --q ! !C MM'1 % %E}}Z)ABBHF##   g & &      !"   G&+,7,,G }}XGWb}QQH  ]]__F ;++I-- r22Ka]]<AA V+VV*VVWWW 62&&**7B77I DBCCC-- **K k4 A   s7II Ictjdd}|std|di}|dt}|dpt }|drd}n2|d rd }n|d rd }nd }t} ||5} | jj ||||} tj | j } dddn #1swxYwYn^#t$rt$rG} td| dt#dt%| j| d} ~ wwxYwt)|d5} | | dddn #1swxYwY|S)zGenerate audio using Mistral Voxtral TTS API. The API returns base64-encoded audio; this function decodes it and writes the raw bytes to *output_path*. Supports native Opus output for Telegram voice bubbles. MISTRAL_API_KEYrz?MISTRAL_API_KEY not set. Get one at https://console.mistral.ai/mistralrrr\rrrrrrr)rrrrNzMistral TTS failed: %sTr@zMistral TTS failed: r)rvrrrCDEFAULT_MISTRAL_TTS_MODELDEFAULT_MISTRAL_TTS_VOICE_IDrr#rrcompletebase64 b64decode audio_datarGrEerrorrtype__name__rr)rrrLr mi_configrrrr!rrrrJrs r_generate_mistral_ttsrsi)2..G \Z[[[y"--I MM'#< = =E}}Z((H,HHF##     f % %    g & &  $&&G M WW % % % @|*33! / 4H !*8+>??K @ @ @ @ @ @ @ @ @ @ @ @ @ @ @  MMM -q4 @@@D$q''2BDDEE1LM k4 A   sO D.=D" D."D&&D.)D&*D..F AFF F>>GG pcm_bytesrchannels sample_widthc @ddl}||z|z}||z}t|}|dddd|||||dz }|dd |} d t|zt| z|z} |d d | d } | |z| z|zS)aWrap raw signed-little-endian PCM with a standard WAV RIFF header. Gemini TTS returns audio/L16;codec=pcm;rate=24000 -- raw PCM samples with no container. We add a minimal WAV header so the file is playable and ffmpeg can re-encode it to MP3/Opus downstream. rNz <4sIHHIIHHsfmt r3z<4sIsdataz<4sI4ssRIFFsWAVE)structlenpack) rrrrr byte_rate block_align data_size fmt_chunkdata_chunk_header riff_size riff_headers r_wrap_pcm_as_wavr(sMMMh&5I\)KII   q  I FGY??C NN"S):%;%;;iGI++hGDDK  "%6 6 BBrc ddl}tjdptjdpd}|st d|di}t |dtpt}t |d tpt}t |d ptjd pt d }d d|igigdgddd|iiidd} |d|d} | | d|iddi| d} | j dkr | di} | dp| jdd} n#t$r| jdd} YnwxYwt!d | j d!|  | }|d"dd#d }t#d$|Dd}|t!d%|d&p|d'pi}|d(d}n2#t$t&t(f$r}t!d)||d}~wwxYw|st!d*t+j|}t/|}|d+r?t5|d,5}||dddn #1swxYwY|St9jd+d-.5}|||j}dddn #1swxYwY t?j d/}|r|d0r|d1|d2d3d4d5d6d7d8d9d:d;d|g}n |d1|d:d;d|g}tCj"|d}|j#dkr6|j$%d?d@Add}t!dB|n0tL'dC|t?j(|| tj)|n:#tT$rYn.wxYw# tj)|w#tT$rYwwxYwxYw|S)DaGenerate audio using Google Gemini TTS. Gemini's generateContent endpoint with responseModalities=["AUDIO"] returns raw 24kHz mono 16-bit PCM (L16) as base64. We wrap it with a WAV RIFF header to produce a playable file, then ffmpeg-convert to MP3 / Opus if the caller requested those formats (same pattern as NeuTTS). Args: text: Text to convert (prompt-style; supports inline direction like "Say cheerfully:" and audio tags like [whispers]). output_path: Where to save the audio file (.wav, .mp3, or .ogg). tts_config: TTS config dict. Returns: Path to the saved audio file. rNGEMINI_API_KEYGOOGLE_API_KEYrzIGEMINI_API_KEY not set. Get one at https://aistudio.google.com/app/apikeygeminirrrGEMINI_BASE_URLrpartsrAUDIO voiceConfigprebuiltVoiceConfig voiceName)responseModalities speechConfig)contentsgenerationConfigz/models/z:generateContentkeyrrr)paramsrrrjrormessagei,zGemini TTS API error (HTTP r candidatesrc3*K|]}d|vsd|v |VdS) inlineData inline_dataNrW).0ps r z'_generate_gemini_tts..ns7WW|q/@/@MUVDVDV1DVDVDVDVWWrz+Gemini TTS response contained no audio datar<r=rz#Gemini TTS response was malformed: z$Gemini TTS returned empty audio datarrFsuffixdeleterTr\r]r^r_r`rarbrcrdrerf -loglevelTrgrhrkrlrmzffmpeg conversion failed: zEffmpeg not found; writing raw WAV to %s (extension may be misleading))+rrvrrQrrCr;DEFAULT_GEMINI_TTS_MODELDEFAULT_GEMINI_TTS_VOICEDEFAULT_GEMINI_TTS_BASE_URLrrrrrrGrnextKeyError IndexError TypeErrorrrr(rPrrrtempfileNamedTemporaryFilenamerUrVrqrrrsrtrurErHcopyfileremoveOSError)rrrLrr gemini_configrrrrendpointrerrdetailrr. audio_partinline audio_b64rJr wav_bytesrtmpwav_pathrTcmdr}rts r_generate_gemini_ttsr]*s+"OOOy)**Obi8H.I.IORVVXXG   W   NN8R00M  !!'+CDD E E K K M M iQiE  !!'+CDD E E K K M M iQiE*%% ' 9& ' ' ' & eggffSkk  /01#*))K+?     G;;E;;;H}}w!34  Hs"" )--//%%gr22CWWY''>8=#+>FF ) ) )]4C4(FFF ) K(*> K K6 K K    M}}\"1%i09WWeWWWY]^^  LMM M --T 1N1NTRTJJvr** j) ,MMMDDDEE1LM CABBB ++I ++I##F++ +t $ $  GGI                    $F5 A A AS )8h''  3  ""++F33 XD(y%E65+w tXt[';W^CbIIIF A%%--gh-GGM"#H#H#HIII& NNW    OHk 2 2 2  Ih        D   Ih        D  sA G G<;G<BJ66K% K  K%M55M9<M9OOO C!S/S SSS;S+*S;+ S85S;7S88S;cf ddl}|jdduS#t$rYdSwxYw)z=Check if the neutts engine is importable (installed locally).rNneuttsF)importlib.utilutil find_specrG) importlibs r_check_neutts_availablerdsP~''11== uus " 00cZtttjdz dz S)z9Return path to the bundled default voice reference audio.neutts_sampleszjo.wavr;r__file__parentrWrr_default_neutts_ref_audiorj$ tH~~$'77(B C CCrcZtttjdz dz S)z>Return path to the bundled default voice reference transcript.rfzjo.txtrgrWrr_default_neutts_ref_textrmrkrcddl}|di}|ddp t}|ddp t}|dd}|d d }|} |d s|d d dd z} t ttj dz } |j | d|d| d|d|d|d|g} tj | ddd} | j dkrk| j} d| D}t#dt%d|pd| |kr`t)jd}|r5|d| ddd |g}tj |dd!"t-j| nt-j| ||S)#aGenerate speech using the local NeuTTS engine. Runs synthesis in a subprocess via tools/neutts_synth.py to keep the ~500MB model in a separate process that exits after synthesis. Outputs WAV; the caller handles conversion for Telegram if needed. rNr_ ref_audiorref_textrzneuphonic/neutts-air-q4-ggufdevicecpurr[r3zneutts_synth.pyz--textz--outz --ref-audioz --ref-textz--modelz--deviceTx)rirrjc<g|]}|d|S)zOK:) startswith)r>ls r z$_generate_neutts..s)QQQQQ\\%=P=PQqQQQrzNeuTTS synthesis failed: rrTr]rfrDrrg)checkrj)sysrCrjrmrrpr;rrhri executablerqrrrsrtrQ splitlinesrchrjoinrUrVrvrPrename)rrrLrz neutts_configrorprrqr[ synth_scriptr\r}rt error_linesrTconv_cmds r_generate_neuttsrsJJJNN8R00M!!+r22Q6O6Q6QI  R00N4L4N4NH   g'E F FE   x / /FH    ' ':%%c1--a069tH~~,/@@AAL  $yh5F C^C4 M M MF A$$&&QQ&"3"3"5"5QQQ es2ww||K7P7P7cTceefff;h''  -hk7KXH N84 < < < < Ih     Ih , , , rc rstddSttkr=tdttdtt t}ddlm }|dd  }|d k}|r"t| }nutj d }tt }|d d |r|dvr |d|dz }n |d|dz }|jd d t'| |dkrf t)n)#t*$rt-jddddcYSwxYwtdt3n|dkrf t5n)#t*$rt-jddddcYSwxYwtdt7n|dkr-tdt9n[|dkr-tdt;n(|dkrf t=n)#t*$rt-jddddcYSwxYwtd t?n|d!kr-td"tAn|d#krTtCst-jdd$ddStd%tEn/d } tGn#t*$rd} YnwxYw| rtd& ddl$} | j%&d'(5} | 'fd)(d*+dddn #1swxYwYn#tR$r&tUj+tYYnYwxYwtCr.td,d#}tEnt-jdd-ddStZj./r#tZj.0dkrt-jdd.|d/ddSd} |d0vr+1dste} | r| d } n|d1vr1d} tZj.0}td2|d3|d4}| rd5|}t-jd ||| d6dS#tf$r>}d7|d8|}t4d9|t|dcYd}~Sd}~wtj$r@}d:|d8|}t4d9|d ;t|dcYd}~Sd}~wtl$r@}d<|d8|}t4d9|d ;t|dcYd}~Sd}~wwxYw)=ac Convert text to speech audio. Reads provider/voice config from ~/.hermes/config.yaml (tts: section). The model sends text; the user configures voice and provider. On messaging platforms, the returned MEDIA: tag is intercepted by the send pipeline and delivered as a native voice message. In CLI mode, the file is saved to ~/voice-memos/. Args: text: The text to convert to speech. output_path: Optional custom save path. Defaults to ~/voice-memos/.mp3 Returns: str: JSON result with success, file_path, and optionally MEDIA tag. zText is requiredF)successz.TTS text too long (%d chars), truncating to %dNr)get_session_envHERMES_SESSION_PLATFORMrtelegramz %Y%m%d_%H%M%ST)parentsexist_ok)rrr r,tts_r\z.mp3rz`ElevenLabs provider selected but 'elevenlabs' package not installed. Run: pip install elevenlabs)rr) ensure_asciiz$Generating speech with ElevenLabs...rzz%text_to_speech_tool..~sGK0B4S]0^0^$_$_rrrjz9Edge TTS not available, falling back to NeuTTS (local)...zhNo TTS provider available. Install edge-tts (pip install edge-tts) or set up NeuTTS for local synthesis.z-TTS generation produced no output (provider: ))r(r_rr)rrr r,z,TTS audio saved: %s (%s bytes, provider: %s),zMEDIA:z[[audio_as_voice]] )r file_path media_tagrNvoice_compatiblezTTS configuration error (rz%szTTS dependency missing (r@zTTS generation failed ()7rQ tool_errorrMAX_TEXT_LENGTHrErHrKrRgateway.session_contextrrPr expanduserdatetimenowstrftimeDEFAULT_OUTPUT_DIRmkdirrir;rrDrdumpsinforrrrrr#rr]rdrrconcurrent.futuresfuturesThreadPoolExecutorsubmitr}rrrrrrvrwrxryrr~rrr{rG)rrrNrplatform want_opusr timestampout_diredge_available concurrentpoolr opus_path file_sizerrJ error_msgrrLs` @@rtext_to_speech_toolrs* =tzz||=,e<<<< 4yy?""GTTcddd$_$%!##JZ((H 8777778"==CCEEHZ'I 9%%0022 %))++44_EE )** dT 222  9%RRR"8"8"8"88II"8"8"8"88I4$7779~~HG4 | # # '"$$$$ ' ' 'z$##!&'''''' ' KK> ? ? ? x < < < <  ! ! '%'''' ' ' 'z$[##!&'''''' ' KK> ? ? ? x < < < <  " " KK? @ @ @ !$* = = = =    KK; < < < dHj 9 9 9 9  " " '&(((( ' ' 'z$H##!& '''''' ' KKG H H H !$* = = = =  ! ! KKE F F F x < < < <  ! !*,, 'z$F##!& '''' KKB C C C T8Z 8 8 8 8"N ' """" ' ' '!& ' ' @AAAP----#+>>1>MM-QU ______ &&,,,---------------$PPPK 24: N NOOOOOP(** ' WXXX# x<<<<z$E##!& ''''w~~h'' #27??8+D+D+I+I: TTTT"### #! ; ; ;HDUDUV\D]D] ;(22I ($#' F F F'0088 GOOH--  BHQZN^N^`hiii(X''  ;:y::Iz!" 0      444@@@Q@@  T9%%%)U333333333 444?x??A??  T9t 444)U333333333 444>h>>1>>  T9t 444)U333333333 4s1 XF"!X"#GXG6X?H X#H41X3H44BXK X #LXLBX/XOX O$!X#O$$XQ*#/Q Q*Q""Q*%Q"&Q*)X*-RXRAX2AXB/X [ 3Y>[ [5Z [ [5[[[cL tdS#t$rYnwxYw ttjdrdSn#t$rYnwxYw t t rdSn#t$rYnwxYwtjdrdStjdrdStjdstjdrdS ttjdrdSn#t$rYnwxYwtrdSdS) z Check if at least one TTS provider is available. Edge TTS needs no API key and is the default, so if the package is installed, TTS is available. Returns: bool: True if at least one provider can work. Trrrr*r+r F) rrDrrvrr_has_openai_audio_backendr#rdrWrrcheck_tts_requirementsrs t        9) * * 4         $ & & 4        y"##t yt y!""bi0@&A&At    9& ' ' 4         t 5sA "A AAA88 BB"D DDc t}|rtds |tfStd}|$d}t r|dz }t ||jt|j dddfS)zReturn direct OpenAI audio config or a managed gateway fallback. When ``tts.use_gateway`` is set in config, the Tool Gateway is preferred even if direct OpenAI credentials are present. r? openai-audioNz8Neither VOICE_TOOLS_OPENAI_KEY nor OPENAI_API_KEY is setz5, and the managed OpenAI audio gateway is unavailablerv1) r r DEFAULT_OPENAI_BASE_URLr r rnous_user_tokenrgateway_originr)direct_api_keymanaged_gatewayr9s rrrs 233N7oe4476662>BBOL % ' ' O N NG!!!  *G  ) 0 0 5 5888$-- rcVttptdS)zPReturn True when OpenAI audio can use direct credentials or the managed gateway.r)boolr r rWrrrrs% ,..^2N~2^2^ _ __rz(?<=[.!?])(?:\s|\n)|(?:\n\n)z```[\s\S]*?```z\[([^\]]+)\]\([^)]+\)z https?://\S+z \*\*(.+?)\*\*z \*(.+?)\*z`(.+?)`z^#+\s*flagsz ^\s*[-*]\s+z---+z\n{3,}cFtd|}td|}td|}td|}t d|}t d|}td|}td|}td|}td|}| S)z:Remove markdown formatting that shouldn't be spoken aloud. z\1rz ) _MD_CODE_BLOCKsub_MD_LINK_MD_URL_MD_BOLD _MD_ITALIC_MD_INLINE_CODE _MD_HEADER _MD_LIST_ITEM_MD_HR _MD_EXCESS_NLrQ)rs r_strip_markdown_for_ttsrs   c4 ( (D <<t $ $D ;;r4 D <<t $ $D >>% & &D   ud + +D >>"d # #D   R & &D ::b$  D   VT * *D ::<<r text_queue stop_eventtts_done_eventdisplay_callbackc | ddttt}|di}|d|d|dt jdd}|stdn t}|| n*#t$rtd YnwxYw t}| d d d  nj#ttf$r'} td| dYd} ~ n7d} ~ wt $r'} td| dYd} ~ nd} ~ wwxYwd} d} d} d} gt#jdt"j}dt(ffd }ds; || }n5#t,j$r#t1| | kr || d} Y^wxYw|6|d| } | r || n| |z } |d| } d| vrd| vr t6| }|n_|}| d|}| |d} t1|| kr|| z} n |||; |n#t,j$rYnwxYw,n2#t $r%} td| Yd} ~ nd} ~ wwxYw:  n#t $rYnwxYw|!dS#:  n#t $rYnwxYw|!wxYw)aConsume text deltas from *text_queue*, buffer them into sentences, and stream each sentence through ElevenLabs TTS to the speaker in real-time. Protocol: * The producer puts ``str`` deltas onto *text_queue*. * A ``None`` sentinel signals end-of-text (flush remaining buffer). * *stop_event* can be set to abort early (e.g. user interrupt). * *tts_done_event* is **set** in the ``finally`` block so callers waiting on it (continuous voice mode) know playback is finished. Nrrstreaming_model_idrrrz8ELEVENLABS_API_KEY not set; streaming TTS audio disabledrz8elevenlabs package not installed; streaming TTS disabledr0r3int16) sampleraterdtypezsounddevice not available: %sz#sounddevice OutputStream failed: %srg?z].*?rsentencec\rdSt|}|sdS|d} D]0}|d|krdS1 |  | dSt |tkr|dt} j | d}h|D]a}rnLddl }| ||j } |ddbdSdS |dS#t$r&}t d |Yd}~dSd}~wwxYw) z6Display sentence and optionally generate + play audio.Nz.!, pcm_24000rr)rrr3z!Streaming TTS sentence failed: %s)is_setrrQrPrappendrrrrnumpy frombufferrrreshaperGrErH)rcleaned cleaned_lowerprev audio_iterr_np audio_arrayexc_play_via_tempfile_spoken_sentencesrrr output_streamrrs r_speak_sentencez.stream_tts_to_speaker.._speak_sentenceos  "" -h77==??G #MMOO22599M)  ::<<&&u-->>FF?  $ $W - - -+  ***~7||o--!"2?"23 I#2:: %%"- ; !,!+HH%,,.."!E++++&)nnU#)n&L&L %++K,?,?A,F,FGGGG HH!E '&z:>>>>> I I IBCHHHHHHHHH Is%BE;- E;; F+F&&F+cd} ddl}tjdd}|j}||d5}|d|d|d |D]-}|rn| |.dddn #1swxYwYdd l m }||n2#t$r%}td |Yd}~nd}~wwxYw|r( tj|dS#t"$rYdSwxYwdS#|r& tj|w#t"$rYwwxYwwxYw) z0Write PCM chunks to a temp WAV file and play it.NrrFrArr3r4r0)play_audio_filez!Temp-file TTS fallback failed: %s)waverLrMrNr setnchannels setsampwidth setframerater writeframestools.voice_moderrGrErHrvunlinkrQ) rstop_evttmp_pathrrZwfrrrs rrz1stream_tts_to_speaker.._play_via_tempfilesH  1NNN8YYsD)).ROOA&&&OOA&&&OOE***!+..#??,,"!Eu----...............=<<<<<)))) I I IBCHHHHHHHH I (+++++"8 (++++"s7CA0B7+ C7B;;C>B;?CD2 DC>9D2>DD2 D D-,D-2E6E  E EEEErzTz Streaming TTS pipeline error: %s)"clearr%DEFAULT_ELEVENLABS_STREAMING_MODEL_IDrKrCrvrrErHrrDr' OutputStreamstartrQrFrGrecompileDOTALLr;rqueueEmptyrrrQ_SENTENCE_BOUNDARY_REsearchend get_nowaitstoprset)rrrrrLrrrr&r sentence_bufmin_sentence_lenlong_flush_len queue_timeout_think_block_rerdeltamend_posrrrrrrrs ` ` @@@@@@rstream_tts_to_speakerr,s"s .8%'' NN<44 ==X66==!5!*z8!D!DFF)0"55 ) NNU V V V V [/11 #G444 [ [ [YZZZZZ [ ! ),..B$&OO#(1G%4%%M"''))))#W-)))LL!@#FFF$(MMMMMM )))NN#H#NNN$(MMMMMM)  ')*%>biPPP( Ic( I( I( I( I( I( I( I( I( I( I( I( I( IT   4##%%* * "}==;   |$$~55#OL111#%L  }.222|DD %%''2#OL111 E !L +..r<@@L<''Jl,J,J *)00>>9%%'''1+GHH5 x~~''((+;;;#+l#:L))) *?##%%* *Z  %%'''';      @@@93????????@  $ ""$$$##%%%%      $ ""$$$##%%%%    sB(M3C"!M3"$D M3D  M3:E M3 F1E=8M3= F1 F,'M3,F11AM3 H"!M3"/IM3IC2M3MM3M.+M3-M..M32O73 N"=NO7N""O7((O OO7Q ;(P$#Q $ P1.Q 0P11Q __main__u🔊 Text-to-Speech Tool Modulez2==================================================c> |dS#t$rYdSwxYw)NTF)rD)importerlabels r_checkrs9  HJJJ4   55 s  z Provider availability:z Edge TTS: installedz$not installed (pip install edge-tts)z ElevenLabs: elz¬ installed (pip install elevenlabs)z API Key: rrznot setz OpenAI: oaiz not installedz2not set (VOICE_TOOLS_OPENAI_KEY or OPENAI_API_KEY)z MiniMax: rz API key setznot set (MINIMAX_API_KEY)z ffmpeg: u ✅ foundu(❌ not found (needed for Telegram Opus)z Output dir: z Configured provider: )registryrraConvert text to speech audio. Returns a MEDIA: path that the platform delivers as a voice message. On Telegram it plays as a voice bubble, on Discord/WhatsApp as an audio attachment. In CLI mode, saves to ~/voice-memos/. Voice and provider are user-configured, not model-selected.objectstringz:The text to convert to speech. Keep under 4000 characters.)r descriptionz9Optional custom file path to save the audio. Defaults to z/audio_cache/.mp3rr)r propertiesrequired)rNr parametersr?c rt|dd|dS)Nrrrr)rrC)argskws rrr1s62 XXfb ! !HH]++ - - -ru🔊)rNtoolsetschemahandlercheck_fnemojir)x__doc__rrrrloggingrvrrrUrqrL threadingrpathlibrtypingrrrr urllib.parserr:r getLoggerrrEtools.managed_tool_gatewayr tools.tool_backend_helpersr r r tools.xai_httprrrrr#r'rOrrrrrrrrrrr r rrrrrrErFrGGEMINI_TTS_SAMPLE_RATEGEMINI_TTS_CHANNELSGEMINI_TTS_SAMPLE_WIDTHr;r<rrrKrRrrXr~rrrrrrrrr(r]rdrjrmrrrtuplerrrrrrrrrr MULTILINErrrrrQueueEventrprintrrrIrNtools.registryrr TTS_SCHEMAregisterrWrrr?s9 2    000000000000 000000  8 $ $CCCCCCpppppppppp000000    '46(;%(5'2=3E,9!P=====-,.. $sCx.&Ld38nLLLLL.T.... s x}    L3Sd3PS8nY\<(s(($sCx.(UX((((\-s--$sCx.-UX----f;C;c;tCH~;RU;;;;BQQ#Q4S>QVYQQQQn++#+4S>+VY++++f.'/ CCCCC C  CCCCD@s@@$sCx.@UX@@@@ND3DDDD D#DDDD 232S2d38n2QT2222t"&A4A4 A4#A4 A4A4A4A4N)))))XU38_,`4````# #BCC-.. 2:. / / "*_ % % 2:& ' ' RZ % % "*Z(( RZ  6 6 6  >>>> G   9%%  # #    &9= FF FFOFxt 45 FFFFX z E +,,, E(OOO E $%%% E x&&1A6*J*Jv;;Pv x xyyy E z&&1CT*J*Jx;;Px z z{{{ E TIBI.B$C$CR55 T TUUU E e&&1F*N*Nc;;Tc e efff E o0022 l558l o o E kIBI6G,H,Hi==Ni k klll E i++--g;;=g i ijjj E 1/ 1 1222    F}V$$H E .H . ./// 0/////// n![ ! O[n[n[p[p O O O   H   &    - -$       r