ei^ UdZddlZddlZddlZddlZddlZddlZddlZddlZddl Z ddl Z ddl Z ddl Z ddl Z ddlmZddlmZmZmZmZddlmZddlmZejeZddlmZddlmZm Z m!Z!dd l"m#Z#d Z$d Z%d Z&d Z'dZ(dZ)dZ*dZ+dZ,dZ-dZ.dZ/dZ0dZ1dZ2dZ3dZ4dZ5dZ6dZ7dZ8dZ9d Z:d!Z;d"Zd%Z?d&Z@d!ZAd'ZBd(ZCd)eDfd*ZEeEZFd+d,d-d.d/d+d.d0d0d1 ZGeeDeHfeId2<d+d+d.d.d.d.d3d4d5ZJeeDeHfeId6<d/ZKeKZL dd7eeDd8eeeDefd)eHfd9ZMd)eeDeffd:ZNd8eeDefd)eDfd;ZOd)ePfd<ZQd=eDd)eeDfd>ZRd?eDd@eDd8eeDefd)eDfdAZSd?eDd@eDd8eeDefd)eDfdBZTd?eDd@eDd8eeDefd)eDfdCZUd?eDd@eDd8eeDefd)eDfdDZVd?eDd@eDd8eeDefd)eDfdEZWd?eDd@eDd8eeDefd)eDfdFZXeAeBeCfdGeYdHeHdIeHdJeHd)eYf dKZZd?eDd@eDd8eeDefd)eDfdLZ[d)ePfdMZ\d)ePfdNZ]d)eDfdOZ^d)eDfdPZ_d?eDd@eDd8eeDefd)eDfdQZ`iaaeeDefeIdR<d?eDd@eDd8eeDefd)eDfdSZb dd?eDd@eeDd)eDfdTZcd)ePfdUZdd)eeeDeDffdVZfd)ePfdWZgejhdXZiejhdYZjejhdZZkejhd[Zlejhd\Zmejhd]Znejhd^Zoejhd_ejp`Zqejhdaejp`ZrejhdbZsejhdcZtd?eDd)eDfddZu ddeejvdfe jwdge jwdheeeDgdffdiZxedjkreydkeydldmZzeydneydoeze$drdpndqeydreze%dsrdpndteyduej{dvrdwndxeydyeze&dzrdpnd{eydue!rdwnd|eyd}ej{d~rdndeydeQrdndeydeFeNZ|eOe|Z}eyde}ddl~mZmZddddddddedddd?gddZejddededddS)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 0/home/div/.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 Ircddlm}|S)z?Lazy import KittenTTS. Returns the class or raises ImportError.r KittenTTS) kittenttsr*r)s r_import_kittenttsr,Rs###### redgezen-US-AriaNeuralpNInz6obpgDQGcFmaJgBeleven_multilingual_v2eleven_flash_v2_5zgpt-4o-mini-ttsz!KittenML/kitten-tts-nano-0.8-int8Jasperalloyzhttps://api.openai.com/v1z speech-2.8-hdEnglish_Graceful_Ladyz https://api.minimax.io/v1/t2a_v2zvoxtral-mini-tts-2603z$c69964a6-ab8b-4f8a-9465-ec0925096ec8eveen]zhttps://api.x.ai/v1zgemini-2.5-flash-preview-ttsKorez0https://generativelanguage.googleapis.com/v1betareturnc@ddlm}t|ddS)Nrget_hermes_dirz cache/audio audio_cache)hermes_constantsr>strr=s r_get_default_output_dirrBws.////// ~~m];; < <.max_text_length`` (user override in config.yaml) 2. ElevenLabs model-aware table (keyed on configured ``model_id``) 3. ``PROVIDER_MAX_TEXT_LENGTH`` default 4. ``FALLBACK_MAX_TEXT_LENGTH`` (4000) Non-positive or non-integer overrides fall through to the default so a broken config can't accidentally disable truncation entirely. max_text_lengthNrrGmodel_id) FALLBACK_MAX_TEXT_LENGTHlowerstrip isinstancegetdictboolintDEFAULT_ELEVENLABS_MODEL_IDrPrArI)rQrRkeycfgprov_cfgoverriderUmappeds r_resolve_max_text_lengthrds& ('' ..   " "C  C)#''#,,==Eswws|||2H2:Dx||-...H(D!!(C  X\\ lN'' 33R7R155c(mm6I6I6K6KLL  M # ' '-E F FFrc  ddlm}|}|diS#t$rtdicYSt $r)}td|dicYd}~Sd}~wwxYw) z Load TTS configuration from ~/.hermes/config.yaml. Returns a dict with provider settings. Falls back to defaults for any missing fields. r) load_configttsz9hermes_cli.config not available, using default TTS configzFailed to load TTS config: %sTexc_infoN)hermes_cli.configrfrZ ImportErrorloggerdebug Exceptionwarning)rfconfiges r_load_tts_configrrs 111111zz%$$$  PQQQ 6DIII s!%(&B BA=7B=Bc|dptS)z%Get the configured TTS provider name.rQ)rZDEFAULT_PROVIDERrWrX)rRs r _get_providerrus2 NN: & & :*: A A C C I I K KKrc.tjdduS)z+Check if ffmpeg is available on the system.ffmpegN)shutilwhichrr _has_ffmpegr{s < ! ! --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.r9r.oggrw-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: %srh)r{rsplit subprocessrun returncoderlrostderrdecodeospathexistsgetsizeTimeoutExpiredFileNotFoundErrorrn)r|ogg_pathresultrqs r_convert_to_opusrs ==tsA&&q)F2HM tXy) Cx G      ! ! NNM +V]-A-A'RZ-A-[-[\`]`\`-a c c c4 7>>( # # (A(AA(E(EO  $DDDBCCCCC 333122222 MMM91tLLLLLLLLM 4s+A2C*%AC**)E*#E*: E*E%%E*text output_pathc Kt}|di}|dt}t|d|dd}d|i}|dkrt |dz dz}|dd|d<|j|fi|} | |d {V|S) z Generate audio using Edge TTS. Args: text: Text to convert. output_path: Where to save the MP3 file. tts_config: TTS config dict. Returns: Path to the saved audio file. r-voicespeed?dz+d%rateN)rrZDEFAULT_EDGE_VOICEfloatround Communicatesave) rrrR _edge_tts edge_configrrkwargspct communicates r_generate_edge_ttsrs!""I..,,K OOG%7 8 8E +//':>>'3+G+GHH I IEu F ||US[C'((v')'7777K   ; ' '''''''' rctjdd}|std|di}|dt}|dt }|drd}nd }t}|| } | j |||| } t|d 5} | D]} | |  d d d n #1swxYwY|S)z Generate audio using ElevenLabs. Args: text: Text to convert. output_path: Where to save the audio file. tts_config: TTS config dict. Returns: Path to the saved audio file. ELEVENLABS_API_KEYz=ELEVENLABS_API_KEY not set. Get one at https://elevenlabs.io/rGvoice_idrUr opus_48000_64 mp3_44100_128api_keyrrrU output_formatwbN) rgetenv ValueErrorrZDEFAULT_ELEVENLABS_VOICE_IDr^endswithrtext_to_speechconvertopenwrite) rrrRr el_configrrUrrclientaudio_generatorfchunks r_generate_elevenlabsr/sdi,b11G ZXYYY|R00I}}Z)DEEH}}Z)DEEHF##(' ' #%%J Z ( ( (F+33 # 4O k4 A$  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_urlrrropusmp3)rrzx-idempotency-key)rrinputresponse_format extra_headersg?g@closeNrz)#_resolve_openai_audio_client_configrZDEFAULT_OPENAI_MODELDEFAULT_OPENAI_VOICErrrr[rAuuiduuid4maxminaudiospeechcreatestream_to_filegetattrcallable)rrrRrr 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/rCrlanguage sample_ratebit_rater XAI_BASE_URL/.wavwavr)rrrcodecrz/ttsBearer application/json) Authorization Content-Typez User-Agent<)headersjsonrr)requestsrrrXrrZrADEFAULT_XAI_VOICE_IDDEFAULT_XAI_LANGUAGEr]DEFAULT_XAI_SAMPLE_RATEDEFAULT_XAI_BIT_RATEDEFAULT_XAI_BASE_URLrstriprpostrraise_for_statusrrcontent)rrrRrr xai_configrrrrrrpayloadrrrs r_generate_xai_ttsrsOOOi 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$||d%id&d}|std't|}t|d(5}||dddn #1swxYwY|S))a Generate audio using MiniMax TTS API. MiniMax returns hex-encoded audio data. Supports streaming (SSE) and non-streaming modes. This implementation uses non-streaming for simplicity. Args: text: Text to convert (max 10,000 characters). output_path: Where to save the audio file. tts_config: TTS config dict. Returns: Path to the saved audio file. rNMINIMAX_API_KEYrz@MINIMAX_API_KEY not set. Get one at https://platform.minimax.io/rDrrrr9volpitchrrr.flacflacrF)rrr r i}r7)rbitrateformatchannel)rrstream voice_setting audio_settingrr)rrr)rrr base_resp status_code status_msg unknown errorzMiniMax TTS API error (code ): datarz%MiniMax TTS returned empty audio datar)rrrrrZDEFAULT_MINIMAX_MODELDEFAULT_MINIMAX_VOICE_IDDEFAULT_MINIMAX_BASE_URLrrrr RuntimeErrorbytesfromhexrr)rrrRrr mm_configrrrr r r audio_formatrrrrrrr hex_audio audio_bytesrs r_generate_minimax_ttsr$sOOOi)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/rErrrrrrr r rr)rrrrNzMistral TTS failed: %sTrhzMistral TTS failed: r)rrrrZDEFAULT_MISTRAL_TTS_MODELDEFAULT_MISTRAL_TTS_VOICE_IDrr#rrcompletebase64 b64decode audio_datarnrlerrorrtype__name__rr)rrrRr mi_configrrrr!rrr#rqrs r_generate_mistral_ttsr1(si)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 r9z<4sIsdataz<4sI4ssRIFFsWAVE)structlenpack) r2rr3r4r9 byte_rate block_align data_size fmt_chunkdata_chunk_header riff_size riff_headers r_wrap_pcm_as_wavrCYsMMMh&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$rYwwxYwxYw|S)DaGenerate audio using Google Gemini TTS. Gemini's generateContent endpoint with responseModalities=["AUDIO"] returns raw 24kHz mono 16-bit PCM (L16) as base64. We wrap it with a WAV RIFF header to produce a playable file, then ffmpeg-convert to MP3 / Opus if the caller requested those formats (same pattern as NeuTTS). Args: text: Text to convert (prompt-style; supports inline direction like "Say cheerfully:" and audio tags like [whispers]). output_path: Where to save the audio file (.wav, .mp3, or .ogg). tts_config: TTS config dict. Returns: Path to the saved audio file. rNGEMINI_API_KEYGOOGLE_API_KEYrzIGEMINI_API_KEY not set. Get one at https://aistudio.google.com/app/apikeyrFrrrGEMINI_BASE_URLrpartsrAUDIO voiceConfigprebuiltVoiceConfig voiceName)responseModalities speechConfig)contentsgenerationConfigz/models/z:generateContentr_rrr)paramsrrrrr-messagei,zGemini TTS API error (HTTP r candidatesrc3*K|]}d|vsd|v |VdS) inlineData inline_dataNrz).0ps r z'_generate_gemini_tts..s7WW|q/@/@MUVDVDV1DVDVDVDVWWrz+Gemini TTS response contained no audio datarUrVrz#Gemini TTS response was malformed: z$Gemini TTS returned empty audio datarrFsuffixdeleterwrrrrrrrrrrr -loglevelTrrrrrzffmpeg conversion failed: zEffmpeg not found; writing raw WAV to %s (extension may be misleading))+rrrrXrrZrADEFAULT_GEMINI_TTS_MODELDEFAULT_GEMINI_TTS_VOICEDEFAULT_GEMINI_TTS_BASE_URLrrrrrrnrnextKeyError IndexError TypeErrorr*r+rCrWrrrtempfileNamedTemporaryFilenamerxryrrrrrrlrocopyfileremoveOSError)rrrRrr gemini_configrrrrendpointrerrdetailrrH audio_partinline audio_b64rqr2 wav_bytesrtmpwav_pathrwcmdrrs r_generate_gemini_ttsrv{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).rNrHFimportlib.utilutil find_specrn importlibs r_check_neutts_availabler~sP~''11== uu " 00cf ddl}|jdduS#t$rYdSwxYw)z@Check if the kittentts engine is importable (installed locally).rNr+Frxr|s r_check_kittentts_availabler sP~'' 44D@@ uurcZtttjdz dz S)z9Return path to the bundled default voice reference audio.neutts_sampleszjo.wavrAr__file__parentrzrr_default_neutts_ref_audior$ tH~~$'77(B C CCrcZtttjdz dz S)z>Return path to the bundled default voice reference transcript.rzjo.txtrrzrr_default_neutts_ref_textrrrcddl}|di}|ddp t}|ddp t}|dd}|d d }|} |d s|d d dd z} t ttj dz } |j | d|d| d|d|d|d|g} tj | ddd} | j dkrk| j} d| D}t#dt%d|pd| |kr`t)jd}|r5|d| ddd |g}tj |dd!"t-j| nt-j| ||S)#aGenerate speech using the local NeuTTS engine. Runs synthesis in a subprocess via tools/neutts_synth.py to keep the ~500MB model in a separate process that exits after synthesis. Outputs WAV; the caller handles conversion for Telegram if needed. rNrH ref_audiorref_textrzneuphonic/neutts-air-q4-ggufdevicecpurr~r9zneutts_synth.pyz--textz--outz --ref-audioz --ref-textz--modelz--deviceTx)rrrc<g|]}|d|S)zOK:) startswith)rWls r z$_generate_neutts..Bs)QQQQQ\\%=P=PQqQQQrzNeuTTS synthesis failed: rrwrrr]r-rcheckr)sysrZrrrrrArrr executablerrrrrX splitlinesrchrjoinrxryrrirename)rrrRr neutts_configrrrrrt synth_scriptrurr error_linesrwconv_cmds r_generate_neuttsrsJJJNN8R00M!!+r22Q6O6Q6QI  R00N4L4N4NH   g'E F FE   x / /FH    ' ':%%c1--a069tH~~,/@@AAL  $yh5F C^C4 M M MF A$$&&QQ&"3"3"5"5QQQ es2ww||K7P7P7cTceefff;h''  -hk7KXH N84 < < < < Ih     Ih , , , r_kittentts_model_cachecnt}|di}|dt}|dt}|dd}|dd}|tvrHt d|||t|<t d t|} | |||| } d d l} |} | d s| ddd d z} | | | d| |kr`tj d} | r5| d| ddd|g}tj|ddt!j| nt!j| ||S)atGenerate speech using KittenTTS local ONNX model. KittenTTS is a lightweight TTS engine (25-80MB models) that runs entirely on CPU without requiring a GPU or API key. Args: text: Text to convert to speech. output_path: Where to save the audio file. tts_config: TTS config dict. Returns: Path to the saved audio file. r+rrrr clean_textTz[KittenTTS] Loading model: %sz%[KittenTTS] Model loaded successfully)rrrrNrr~r9r6rwrrr]r-rr)r,rZDEFAULT_KITTENTTS_MODELDEFAULT_KITTENTTS_VOICErrlinfogenerate soundfilerrrrxryrrrrir)rrrRr* kt_config model_namerrrrrsfrtrwrs r_generate_kittenttsr[s"##I{B//Iw(?@@J MM'#: ; ;E MM'3 ' 'E|T22J/// 3Z@@@-6Yz-B-Bz* ;<<< ": .E NN4uEjN Q QEH    ' ':%%c1--a069HHXue$$$;h''  -hk7KXH N84 < < < < Ih     Ih , , , rcrstddStt}t |}t |kr4t d|t |d|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*$rt-jddddcYSwxYwt dt3nf|dkrf t5n)#t*$rt-jddddcYSwxYwt dt7n|dkr-t dt9n|dkr-t dt;n|dkrf t=n)#t*$rt-jddddcYSwxYwt d t?n(|d!kr-t d"tAn|d#krTtCst-jdd$ddSt d%tEn|d&krf tGn)#t*$rt-jdd'ddcYSwxYwt d(tIn/d } tKn#t*$rd} YnwxYw| rt d) ddl&} | j'(d*+5} | )fd,*d-.dddn #1swxYwYn#tV$r&tYj-t]YnYwxYwtCr.t d/d#}tEnt-jdd0ddSt^j01r#t^j02dkrt-jdd1|d2ddSd} |d3vr+3dsti}|r|d } n|d4vr3d} t^j02}t d5|d6|d7}| rd8|}t-jd ||| d9dS#tj$r>}d:|d;|}t 6d<|t|dcYd}~Sd}~wtn$r@}d=|d;|}t 6d<|d >t|dcYd}~Sd}~wtp$r@}d?|d;|}t 6d<|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 for provider %s (%d chars), truncating to %dNr)get_session_envHERMES_SESSION_PLATFORMrtelegramz %Y%m%d_%H%M%ST)parentsexist_ok)rrGrErFtts_rz.mp3rGz`ElevenLabs provider selected but 'elevenlabs' package not installed. Run: pip install elevenlabs)rr-) ensure_asciiz$Generating speech with ElevenLabs...rzz%text_to_speech_tool..)sGK0B4S]0^0^$_$_rrrz9Edge TTS not available, falling back to NeuTTS (local)...zhNo TTS provider available. Install edge-tts (pip install edge-tts) or set up NeuTTS for local synthesis.z-TTS generation produced no output (provider: ))r-rHrDrCr+)rGrrErFz,TTS audio saved: %s (%s bytes, provider: %s),zMEDIA:z[[audio_as_voice]] )r file_path media_tagrQvoice_compatiblezTTS configuration error (rz%szTTS dependency missing (rhzTTS generation failed ()9rX tool_errorrrrurdr:rlrogateway.session_contextrrWr expanduserdatetimenowstrftimeDEFAULT_OUTPUT_DIRmkdirrrArrkrdumpsrrrrr$rr#r1rvr~rr,rrconcurrent.futuresfuturesThreadPoolExecutorsubmitrrrrrrrrrrrrr-rrn)rrrQmax_lenrplatform want_opusr timestampout_diredge_available concurrentpoolr opus_path file_sizerrq error_msgrrRs` @@rtext_to_speech_toolrs * =tzz||=,e<<<<!##JZ((H'x< ? ? ? 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  $ $ '!#### ' ' 'z$J## !& '''''' ' KKL M M M h ; ; ; ;"N ' """" ' ' '!& ' ' @AAAP----#+>>1>MM-QU ______ &&,,,---------------$PPPK 24: N NOOOOOP(** ' WXXX# x<<<<z$E##!& ''''w~~h'' #27??8+D+D+I+I: TTTT"### #! H H HQYQbQbciQjQj H(22I ($#' 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 4sUY/F$#Y/$#G Y/ G  6Y/HY/#H63Y/5H66BY/K"!Y/"#LY/LBY/3Y/ OY/#P>Y/P2Y/4QY/ QY/QY/2S/S  S SSSSY/-TY/TAY/ AY/?B/Y// ]93Z2,]2 ]?5[:4]: ]5]<]]cl tdS#t$rYnwxYw ttjdrdSn#t$rYnwxYw t t rdSn#t$rYnwxYwtjdrdStjdrdStjdstjdrdS ttjdrdSn#t$rYnwxYwtrdStrdSdS) 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. TrrrrErFr&F) rrkrrrr_has_openai_audio_backendr#r~rrzrrcheck_tts_requirementsrns t        9) * * 4         $ & & 4        y"##t yt y!""bi0@&A&At    9& ' ' 4         t!##t 5sA "A AAA88 BB"D DDc t}|rtds |tfStd}|$d}t r|dz }t ||jt|j dddfS)zReturn direct OpenAI audio config or a managed gateway fallback. When ``tts.use_gateway`` is set in config, the Tool Gateway is preferred even if direct OpenAI credentials are present. rg openai-audioNz8Neither VOICE_TOOLS_OPENAI_KEY nor OPENAI_API_KEY is setz5, and the managed OpenAI audio gateway is unavailablerv1) r r DEFAULT_OPENAI_BASE_URLr r rnous_user_tokenrgateway_originr)direct_api_keymanaged_gatewayrRs rrrs 233N7oe4476662>BBOL % ' ' O N NG!!!  *G  ) 0 0 5 5888$-- rcVttptdS)zPReturn True when OpenAI audio can use direct credentials or the managed gateway.r)r\r r rzrrrrs% ,..^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_NLrX)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 di|di|diit jdd}|st dn t}|| n*#t$rt d YnwxYw t}| d d d nj#ttf$r'} td| dYd} ~ n7d} ~ wt"$r'} t d| dYd} ~ nd} ~ wwxYwd} d} d} d} gt%jdt$j}dt*ff d }ds; || }n5#t.j$r#t3| | kr || d} Y^wxYw|6|d| } | r || n| |z } |d| } d| vrd| vr t8| }|n_|}| d|}| |d} t3|| kr|| z} n |||; |n#t.j$rYnwxYw,n2#t"$r%} t d| Yd} ~ nd} ~ wwxYw:  !n#t"$rYnwxYw|"dS#:  !n#t"$rYnwxYw|"wxYw)aConsume text deltas from *text_queue*, buffer them into sentences, and stream each sentence through ElevenLabs TTS to the speaker in real-time. Protocol: * The producer puts ``str`` deltas onto *text_queue*. * A ``None`` sentinel signals end-of-text (flush remaining buffer). * *stop_event* can be set to abort early (e.g. user interrupt). * *tts_done_event* is **set** in the ``finally`` block so callers waiting on it (continuous voice mode) know playback is finished. NrGrstreaming_model_idrUrrz8ELEVENLABS_API_KEY not set; streaming TTS audio disabledrz8elevenlabs package not installed; streaming TTS disabledr6r9int16) samplerater3dtypezsounddevice not available: %sz#sounddevice OutputStream failed: %srg?z].*?rsentencecH rdSt|}|sdS|d} D]0}|d|krdS1 |  | dSt |kr |d} j| d}h|D]a}rnLddl }| ||j } | ddbdSdS |dS#t$r&}td |Yd}~dSd}~wwxYw) z6Display sentence and optionally generate + play audio.Nz.!, pcm_24000rr)rrr9z!Streaming TTS sentence failed: %s)is_setrrXrWrappendr:rrnumpy frombufferrrreshapernrlro)rcleaned cleaned_lowerprev audio_iterr_np audio_arrayexc_play_via_tempfile_spoken_sentencesrrrU output_streamrstream_max_lenrs r_speak_sentencez.stream_tts_to_speaker.._speak_sentence$s  "" -h77==??G #MMOO22599M)  ::<<&&u-->>FF?  $ $W - - -+  ***~7||n,,!/>/2 I#2:: %%"- ; !,!+HH%,,.."!E++++&)nnU#)n&L&L %++K,?,?A,F,FGGGG HH!E '&z:>>>>> I I IBCHHHHHHHHH IsBE1# E11 F!;FF!cd} ddl}tjdd}|j}||d5}|d|d|d |D]-}|rn| |.dddn #1swxYwYdd l m }||n2#t$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.NrrFrZrr9r:r6)play_audio_filez!Temp-file TTS fallback failed: %s)wavererfrgr setnchannels setsampwidth setframerater writeframestools.voice_moderrnrlrorunlinkrj) rstop_evttmp_pathrrswfrrrs rrz1stream_tts_to_speaker.._play_via_tempfileNsH  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_IDrrrZrdrrrlrorrkr' OutputStreamstartrjrmrnrecompileDOTALLrArqueueEmptyr:rrX_SENTENCE_BOUNDARY_REsearchend get_nowaitstoprset)rrrrrRrrrr&r sentence_bufmin_sentence_lenlong_flush_len queue_timeout_think_block_rerdeltamend_posrrrrrUrrrs ` ` @@@@@@@rstream_tts_to_speakerr:s"{ .8%'' NN<44 ==X66==!5!*z8!D!DFF 2  Mz M<)LI)Lz8)L)L M M   )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( I( IT   4##%%* * "}==;   |$$~55#OL111#%L  }.222|DD %%''2#OL111 E !L +..r<@@L<''Jl,J,J *)00>>9%%'''1+GHH5 x~~''((+;;;#+l#:L))) *?##%%* *Z  %%'''';      @@@93????????@  $ ""$$$##%%%%      $ ""$$$##%%%%    sCN$C?>N?$D&#N%D&&N,:E'&N'G8FN G'G N GAN)I?N/I2/N1I22C2N%M:9N:N  N N  NP ON;6P;OP(O// O<;O<Q'(QQ' Q Q'QQ'__main__u🔊 Text-to-Speech Tool Modulez2==================================================c> |dS#t$rYdSwxYw)NTF)rk)importerlabels r_checkr?s9  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: rr1znot 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.objectstringzThe text to convert to speech. Provider-specific character caps apply and are enforced automatically (OpenAI 4096, xAI 15000, MiniMax 10000, ElevenLabs 5k-40k depending on model); over-long input is truncated.)r. descriptionz9Optional custom file path to save the audio. Defaults to z/audio_cache/.mp3rr)r. propertiesrequired)rgrF parametersrgc rt|dd|dS)NrrrrG)rrZ)argskws rrrs62 XXfb ! !HH]++ - - -ru🔊)rgtoolsetschemahandlercheck_fnemojir)__doc__rr*rrloggingrr*r'rxrre threadingrpathlibrtypingrrrr urllib.parserr@r getLoggerr/rltools.managed_tool_gatewayr tools.tool_backend_helpersr r r tools.xai_httprrrrr#r'r,rtrrr^r$rrrrrrrrr'r(rrrrrr^r_r`GEMINI_TTS_SAMPLE_RATEGEMINI_TTS_CHANNELSGEMINI_TTS_SAMPLE_WIDTHrArBrrIr]__annotations__rPrVMAX_TEXT_LENGTHrdrrrur\r{rrrrrr$r1rrCrvr~rrrrrrrrtuplerrr(r,rrrrrr MULTILINErrrrrQueueEventr:printr?rrprQtools.registryrCr TTS_SCHEMAregisterrzrrrjs 2    000000000000 000000  8 $ $CCCCCCpppppppppp000000     '46(;%(="5'2=3E,9!P=====-,..   , ,$sCx.   ##"" 4 4 $sCx.    + ,0"G"Gsm"Gc3h("G "G"G"G"GP$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@@@@NDD3DDDD D#DDDD 232S2d38n2QT2222t*,S#X+++4c44c3h4TW4444x"&S4S4 S4#S4 S4S4S4S4r+++++\U38_,`4````# #BCC-.. 2:. / / "*_ % % 2:& ' ' RZ % % "*Z(( RZ  6 6 6  >>>> G   9%%  # #    &9= NN NNONxt 45 NNNNh 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! s ! O[n[n[p[p O O O   H   &    - -$       r