qix ~UdZddlZddlmZmZddlmZmZmZm Z m Z ddl m Z ddl mZddlmZmZmZmZmZmZmZmZmZmZddlmZdd lmZdd lmZm Z m!Z!m"Z"m#Z#dd l$m%Z%dd l&m'Z'dd l(m)Z)m*Z*ddl+m,Z,ddl-m.Z. ddl/m0Z0ddl1m2Z2ddl3m4Z4de)de9fdZ:edGddZ;e;ddddddde;ddddddde;ddddd ddd!ZGd%d&e'Z?y#e5$r7Z6e jnde6e jnde8de6dZ6[6wwxYw)'a'Sarvam AI Speech-to-Text service implementation. This module provides a streaming Speech-to-Text service using Sarvam AI's WebSocket-based API. It supports real-time transcription with Voice Activity Detection (VAD) and can handle multiple audio formats for Indian language speech recognition. N) dataclassfield)AnyAsyncGeneratorDictLiteralOptional)logger) BaseModel) CancelFrameEndFrame ErrorFrameFrame StartFrameTranscriptionFrameUserStartedSpeakingFrameUserStoppedSpeakingFrameVADUserStartedSpeakingFrameVADUserStoppedSpeakingFrame)FrameDirection) sdk_headers) NOT_GIVEN STTSettings _NotGiven_warn_deprecated_paramis_given)SARVAM_TTFS_P99) STTService)Languageresolve_language)time_now_iso8601) traced_stt) AsyncSarvamAI)ApiError) EventTypez Exception: zEIn order to use Sarvam, you need to `pip install pipecat-ai[sarvam]`.zMissing module: languagereturnctjdtjdtjdtjdtj dtj dtjdtjdtjd tjd tjd tjd i }t||d S)zConvert a Language enum to Sarvam's language code format. Args: language: The Language enum value to convert. Returns: The Sarvam language code string. bn-INgu-INhi-INkn-INml-INmr-INta-INte-INpa-INod-INen-INas-INF) use_base_code)rBN_INGU_INHI_INKN_INML_INMR_INTA_INTE_INPA_INOR_INEN_INAS_INr )r& LANGUAGE_MAPs M/opt/pipecat/venv/lib/python3.12/site-packages/pipecat/services/sarvam/stt.pylanguage_to_sarvam_languagerD:s  L Hl% HHT)frozencdeZdZUdZeed<eed<eed<eeed<eeed<eed<eed<y ) ModelConfigaKImmutable configuration for a Sarvam STT model. Attributes: supports_prompt: Whether the model accepts prompt parameter. supports_mode: Whether the model accepts mode parameter. supports_language: Whether the model accepts language parameter. default_language: Default language code (None = auto-detect). default_mode: Default mode (None = not applicable). use_translate_endpoint: Whether to use speech_to_text_translate_streaming endpoint. use_translate_method: Whether to use translate() method instead of transcribe(). supports_prompt supports_modesupports_languagedefault_language default_modeuse_translate_endpointuse_translate_methodN)__name__ __module__ __qualname____doc__bool__annotations__r strrErCrHrHVs; sm#3-  rErHFunknown)rIrJrKrLrMrNrO transcribe) saarika:v2.5z saaras:v2.5z saaras:v3 MODEL_CONFIGSceZdZUdZedZedzezed<edZ e dzezed<edZ e dzezed <y) SarvamSTTSettingsaFSettings for SarvamSTTService. Parameters: prompt: Optional prompt to guide transcription/translation style/context. Only applicable to models that support prompts (e.g., saaras:v2.5). vad_signals: Enable VAD signals in response. high_vad_sensitivity: Enable high VAD sensitivity. ctSNrrWrErCzSarvamSTTSettings.s9rE)default_factoryNpromptctSr_r`rWrErCrazSarvamSTTSettings.srE vad_signalsctSr_r`rWrErCrazSarvamSTTSettings.sR[rEhigh_vad_sensitivity) rPrQrRrSrrcrVrrUrerTrgrWrErCr]r]sU&+;L%MFC$J "M+0AR+SKy(S49J[4\$+ 1\rEr]ceZdZUdZeZeed<GddeZdddddde ddd d e d e e d e e d d e e de de ede ede ede edeffdZdede fdZde e fdZdefdZdedeffd Zdedee efffd Zde e fd Zdeffd! Zdeffd" Z de!ffd# Z"d$e#de$edffd%Z%d&Z&d'Z'd(Z(d)Z)e* d3d*e d+ede efd,Z+d-e defd.Z,defd/Z-d0e#fd1Z.d2Z/xZ0S)4SarvamSTTServiceaSarvam speech-to-text service. Provides real-time speech recognition using Sarvam's WebSocket API. Event handlers available (in addition to STTService events): - on_connected(service): Connected to Sarvam WebSocket - on_disconnected(service): Disconnected from Sarvam WebSocket - on_connection_error(service, error): Connection error occurred Example:: @stt.event_handler("on_connected") async def on_connected(service): ... _settingsc|eZdZUdZdZeeed<dZee ed<dZ ee ded<dZ ee ed<dZee ed<y) SarvamSTTService.InputParamsaConfiguration parameters for Sarvam STT service. .. deprecated:: 0.0.105 Use ``settings=SarvamSTTSettings(...)`` instead. Parameters: language: Target language for transcription. - saarika:v2.5: Defaults to "unknown" (auto-detect supported) - saaras:v2.5: Not used (auto-detects language) - saaras:v3: Defaults to "unknown" (auto-detect supported) prompt: Optional prompt to guide transcription/translation style/context. Only applicable to saaras:v2.5. Defaults to None. mode: Mode of operation for saaras:v3 models only. Options: transcribe, translate, verbatim, translit, codemix. Defaults to "transcribe" for saaras:v3. vad_signals: Enable VAD signals in response. Defaults to None. high_vad_sensitivity: Enable high VAD (Voice Activity Detection) sensitivity. Defaults to None. Nr&rcrY translateverbatimtranslitcodemixmodererg)rPrQrRrSr&r rrUrcrVrrrrerTrgrWrErC InputParamsrlsX $(,(8$+ $ $`dhw[\]d&* Xd^*/3htn3rErsNwavg@) modelrr sample_rateinput_audio_codecparamssettingsttfs_p99_latencykeepalive_timeoutkeepalive_intervalapi_keyrurrrmrvrwrxryrzr{r|c tddddd} |tdtd|| _|ntdt|s\|j| _|j| _|j |j }|j | _|j| _|| j|| j} | tvr>djttj}td| d|d t| |_| j%|jjstd | d |%|jj std | d | j%|jj"std | d ||jj$}t'|Pd||| | | d| ||_||_||_t1|_t5||j2|_d|_d|_d|_| j r3|j?d|j?d|j?dtAjBd|j2y)aInitialize the Sarvam STT service. Args: api_key: Sarvam API key for authentication. model: Sarvam model to use for transcription. .. deprecated:: 0.0.105 Use ``settings=SarvamSTTSettings(model=...)`` instead. mode: Mode of operation. Options: transcribe, translate, verbatim, translit, codemix. Only applicable to models that support it (e.g., saaras:v3). Defaults to the model's default mode. sample_rate: Audio sample rate. Defaults to 16000 if not specified. input_audio_codec: Audio codec/format of the input file. Defaults to "wav". params: Configuration parameters for Sarvam STT service. .. deprecated:: 0.0.105 Use ``settings=SarvamSTTSettings(...)`` instead. settings: Runtime-updatable settings. When provided alongside deprecated parameters, ``settings`` values take precedence. ttfs_p99_latency: P99 latency from speech end to final transcript in seconds. Override for your deployment. See https://github.com/pipecat-ai/stt-benchmark keepalive_timeout: Seconds of no audio before sending silence to keep the connection alive. None disables keepalive. keepalive_interval: Seconds between idle checks when keepalive is enabled. **kwargs: Additional arguments passed to the parent STTService. rZN)rur&rcrergrurxz, zUnsupported model 'z'. Allowed values: .Model '$' does not support prompt parameter.z"' does not support mode parameter.>' does not support language parameter (auto-detects language).)rvrzr{r|ry)api_subscription_keyheaderson_speech_startedon_speech_stoppedon_utterance_endz)Sarvam STT initialized with SDK headers: rW)"r]rrur&rcrrrerg apply_updater[joinsortedkeys ValueError_configrIrJrKrMsuper__init___api_key_mode_input_audio_codecr _sdk_headersr#_sarvam_client_websocket_context_socket_client _receive_task_register_event_handlerr info)selfr}rurrrvrwrxryrzr{r|kwargsdefault_settingsresolved_modelallowed __class__s rCrzSarvamSTTService.__init__s\- !%     "7,=w G%*  "   "8-> ?,2OO )*0-- ';;*!;;D/5/A/A ,8>8S8S 5    ) )( 3*//  .ii}'9'9'; <=G2>2BBUV]U^^_`a a$^4   " " .t||7S7Sw~&66Z[\ \  DLL$>$>w~&66XYZ Z  $ $ 09W9W.))gh  <<<,,D  #-/1%       #4(M,RVRcRcd"&"!  ' '  ( ()< =  ( ()< =  ( (); < ?@Q@Q?RSTrEr&r'ct|S)zConvert pipecat Language enum to Sarvam's language code. Args: language: The Language enum value to convert. Returns: The Sarvam language code string. )rD)rr&s rClanguage_to_service_languagez-SarvamSTTService.language_to_service_languageLs+844rEc|jjrt|jjS|jjS)zFResolve the current language setting to a Sarvam language code string.)rjr&rDrrLrs rC_get_language_stringz%SarvamSTTService._get_language_stringWs4 >> " ".t~~/F/FG G||,,,rEcy)zCheck if this service can generate processing metrics. Returns: True, as Sarvam service supports metrics generation. TrWrs rCcan_generate_metricsz%SarvamSTTService.can_generate_metrics]s rEframe directioncJKt|||d{|jjsjt |t r|j d{yt |tr0|jr#|jjd{yyyy77I7 w)zProcess incoming frames. Handles VAD frames for TTFB tracking when using Pipecat's VAD instead of Sarvam's built-in VAD. N) r process_framerjre isinstancer_start_metricsrrflush)rrrrs rCrzSarvamSTTService.process_framees g#E9555~~))%!<=))+++E#>?&&--33555'@* 6 ,6s3B#B=B#B>B#B! B#B#!B#deltacKt|jrE|j9|jjs#t d|j j dt|trZt|jrE|j9|jjs#t d|j j dt|1|d{}ddh}|j|zr0|jd{|jd{|j!Dcic] \}}||vs ||}}}|r|j#||S77[7Ecc}}ww)aVApply a settings delta, validate, sync state, and reconnect. Args: delta: A :class:`STTSettings` (or ``SarvamSTTSettings``) delta. Returns: Dict mapping changed field names to their previous values. Raises: ValueError: If a setting is not supported by the current model. Nrrrr&rc)rr&rrKrrjrurr]rcrIr_update_settingsr _disconnect_connectitems _warn_unhandled_updated_settings)rrchangedreconnect_fieldskv unhandledrs rCrz!SarvamSTTService._update_settingsusK ENN #(B<<11 dnn2234// u/ 0& (<<// dnn2233WX077'1 <<>, ,""$ $ $--/ ! !&-mmoSdaBR9RQTS S   1 1) <8 % !SsNCE/E#.E/ E% E/!E'"E/9 E)E) E/%E/'E/)E/rcc:Kddl}|j5|jd|j|jj dt dddd|jjs&|#td|jjdytjd |jjd ||j_|jd{|j!d{y#1swYxYw7)7w) aSet the transcription/translation prompt and reconnect. .. deprecated:: 0.0.104 Use ``STTUpdateSettingsFrame(SarvamSTTSettings(prompt=...))`` instead. Args: prompt: Prompt text to guide transcription/translation style/context. Pass None to clear/disable prompt. Only applicable to models that support prompts. rNalwaysz_.set_prompt() is deprecated. Use STTUpdateSettingsFrame(SarvamSTTSettings(prompt=...)) instead.) stacklevelrrz Updating z prompt.)warningscatch_warnings simplefilterwarnrrPDeprecationWarningrrIrrjrur rrcrr)rrcrs rC set_promptzSarvamSTTService.set_prompts   $ $ &   ! !( + MM>>**+,UU"   ||++! dnn2233WX  i 4 45X>? &   mmo)  & !s<DAD BD-D.DDD DDDctKt||d{|jd{y77w)z~Start the Sarvam STT service. Args: frame: The start frame containing initialization parameters. N)rstartrrrrs rCrzSarvamSTTService.starts3 gmE"""mmo #848688ctKt||d{|jd{y77w)zVStop the Sarvam STT service. Args: frame: The end frame. N)rstoprrs rCrzSarvamSTTService.stops6 gl5!!!    " rctKt||d{|jd{y77w)z[Cancel the Sarvam STT service. Args: frame: The cancel frame. N)rcancelrrs rCrzSarvamSTTService.cancels6 gnU###    $ raudioc(K|jsdy tj|jd}|jj dr |jnd|j}|||j d}|jjr%|jjdi|d{n$|jjdi|d{dy7/7 #t$r}td||Yd}~+d}~wwxYww)zSend audio data to Sarvam for transcription. Args: audio: Raw audio bytes to transcribe. Yields: Frame: None (transcription results come via WebSocket callbacks). Nutf-8audio/rencodingrvzError sending audio to Sarvam: )error exceptionrW) rbase64 b64encodedecoder startswithrvrrOrnrY Exceptionr)rr audio_base64r method_kwargses rCrun_sttzSarvamSTTService.run_stts""J  W!++E299'BL **55h?''d5567 &$#//M||003d))33DmDDD4d))44E}EEE  EE W%DQC#HTUV V V WsSDBC)5C%6$C)C'C)D%C)'C)) D2D D DDcKtjd jjt j d}jj sd|d<jj jj rdnd|d<jjjjrdnd|d<j}|||d <jjrjj|d <jj/jjrjj|d <fd }jjr-|jj j"fi|_n,|jj&j"fi|_j$j)d{_jj\jjrFt-j*d d}t/|r$|jjd{fd}j*j1t2j4|j7j9_j=tj>dy77#t@$r7}d_d_jCd||d{7Yd}~yd}~wtD$r7}d_d_jCd||d{7Yd}~yd}~wwxYww)z.Connect to Sarvam WebSocket API using the SDK.zConnecting to Sarvam)rurvtrue flush_signalNfalsererg language_coderrrcchd ji}|g}d|vr9|j|jDcic]\}}|dk7s ||c}}d}|D]} |di|d|icS|||di|Scc}}w#t$r }|}Yd}~nd}~wwxYw |di|cS#t$r }|}Yd}~Xd}~wwxYw)Nadditional_headersrcrequest_optionsrW)rappendr TypeError) connect_fnrrattemptsrrlast_type_errorattempt_kwargsrrs rC_connect_with_sdk_headersz._connect_with_sdk_headers.s$89J9J"K"8v%OOflln$VdaX QT$VW"&&. ,N,) , ,;  ,#.))!+F++'%W%,*+,,);N;;$,*+,s; A3 A3  A99 BB  BB B1%B,,B1rcFjj|y)z)Wrapper to handle async response handler.N) create_task_handle_message)messagers rC_message_handlerz3SarvamSTTService._connect.._message_handleras  !5!5g!>?rEz Connected to Sarvam successfullyzSarvam API error:  error_msgrzFailed to connect to Sarvam: )#r debugrjrurVrvrergrrrJrrcrIrNr"speech_to_text_translate_streamingconnectrspeech_to_text_streaming __aenter__rgetattrcallableonr%MESSAGEr_receive_task_handlerr_create_keepalive_taskrr$ push_errorr)rconnect_kwargslanguage_stringr prompt_setterrrs` rCrzSarvamSTTService._connects +,j ^--"4#3#34N>>--17~.~~))5:>..:T:TZa}-~~22>"nnAAFw56 #779O*2A/||))djj.D)-v&~~$$0T\\5Q5Q+/>>+@+@x( ,<||22*C''JJRR+$+' +D''@@HH+$+' )-(?(?(J(J(L"LD ~~$$0T\\5Q5Q '(;(;\4 P M*'(=(=>>> @    " "9#4#46F G"&!1!1$2L2L2N!OD   ' ' ) KK: ;+#M ?" S"&D &*D #//.@,DPQ/R R R ^"&D &*D #//.KA3,O[\/] ] ] ^sMG K%K&A6KKAC"CC"C*"C''C**DDcK|jsy |jjd{y7#t$r)}|jd||d{7Yd}~yd}~wwxYww)zHandle incoming messages from Sarvam WebSocket. This task wraps the SDK's start_listening() method which processes messages via the registered event handler callback. NzSarvam receive task error: r)rstart_listeningrr)rrs rCrz&SarvamSTTService._receive_task_handlersd ""  \%%557 7 7 \//.I!,MYZ/[ [ [ \sBA+646A+6 A(A#AA#A+#A((A+c Ktjd| |jdk(r|jj}|jj }tjd|d||dk(r||j d{tjd|jdd{|jtd{|jd{y|d k(rtjd |jd d{|jtd{y|jd k(r1|jj}|jj}|r|j|}n4|j}|r|j|}nt j"}|jd d{|r|j%ry|j'|d|d{|j)t+||j,t/|t1|dr|j3n t5|d{|j7d{yyy77777r7X77757#t8$rB}|j;d||d{7|j=d{7Yd}~yd}~wwxYww)zHandle incoming WebSocket message from Sarvam SDK. Processes transcription data and VAD events from the Sarvam service. Args: message: The parsed response object from Sarvam WebSocket. zReceived response: eventsz VAD Signal: z, Occurred at: START_SPEECHNzUser started speakingr END_SPEECHzUser stopped speakingrdatarTdict)resultzFailed to handle message: r)r rtyper signal_type occured_atr_call_event_handlerbroadcast_framerbroadcast_interruptionr transcriptr_map_language_code_to_enumrrr8strip_handle_transcription push_framer_user_idr!hasattrrrVstop_processing_metricsrrstop_all_metrics) rrsignal timestamprrr&rrs rCrz SarvamSTTService._handle_messages  *7)454 *||x' 11#LL33  |F8?9+NO^+--///LL!89223FGGG../GHHH55777|+LL!89223FGGG../GHHH'$\\44 ' : :  #>>}MH&*&?&?&AO&#'#B#B?#S#+>>../ABBB*"2"2"444ZxPPP//*& MM,.$6=gv6NGLLNTWX_T` 22444;( , 0GH7HH CQ5 *//.H,LXY/Z Z Z'') ) ) *sK1A/J# J  -J#9J :J#JJ#.J/J#3K14/J##J$J#JJ#K1BJ#J,J# J AJ#)J*J#J!J#K1 J# J#J#J#J#J#J#J#J#!J## K.,K)KK)K!K)$K1)K..K1ris_finalc Kyw)zzHandle a transcription result with tracing. This method is decorated with @traced_stt for observability. NrW)rrr"r&s rCrz&SarvamSTTService._handle_transcriptions  srctjtjtjtjtj tj tjtjtjtjtjtjtjd }|j|tjS)z2Map Sarvam language code to pipecat Language enum.) r)r*r+r,r-r.r/r0r1r2zen-USr3r4)rr6r7r8r9r:r;r<r=r>r?EN_USr@rAget)rrmappings rCrz+SarvamSTTService._map_language_code_to_enums}^^^^^^^^^^^^^^^^^^^^^^^^^^ {{=(..99rEc|jduS)z6Check if the Sarvam SDK websocket client is connected.N)rrs rC_is_keepalive_readyz$SarvamSTTService._is_keepalive_readys""$..rEsilencecKtj|jd}|jj dr |jnd|j}|||j d}|j jr%|jjdi|d{y|jjdi|d{y7*7w)zSend silent audio via the Sarvam SDK to keep the connection alive. Args: silence: Silent 16-bit mono PCM audio bytes. rrrNrW) rrrrrrvrrOrrnrY)rr*rrrs rC_send_keepalivez SarvamSTTService._send_keepalives ''077@ &&11(;  # #$1123  " ++ << , ,/$%%//@-@ @ @0$%%00A=A A A A As$B C"C #$CCCCc@K|jd{y7w)z$Start processing metrics collection.N)start_processing_metricsrs rCrzSarvamSTTService._start_metricss++---s r_)1rPrQrRrSr]SettingsrUr rsrrVr rintfloatrrrrrTrrrrrrrrrrrr rr rbytesrrrrrrr"rrr)r,r __classcell__)rs@rCriris"!H  4i4: $ %)!&(,04,;-1$'@U@U} @U  P Q @Uc]@U@U%@U,-@U#5/@U$E?@U"@UD 5X 5# 5-hsm- d66>6 )K)DcN)V!x}!F!!!+!(5(^E4K-H(Tn^`C2\ >*@NR  )- 9A(9K  :::&/T/BUB,.rEri)@rSr dataclassesrrtypingrrrrr logurur pydanticr pipecat.frames.framesr r rrrrrrrr"pipecat.processors.frame_processorrpipecat.services.sarvam._sdkrpipecat.services.settingsrrrrrpipecat.services.stt_latencyrpipecat.services.stt_servicerpipecat.transcriptions.languagerr pipecat.utils.timer!(pipecat.utils.tracing.service_decoratorsr"sarvamair#sarvamai.core.api_errorr$sarvamai.core.eventsr%ModuleNotFoundErrorrrrrVrDrHr[rUr]rirWrErCrEs}(??   >493F/?,&0.I(IsI8 $. "$"#!"!$"') tC$%>  ] ]  ]~ .z~ .Q,FLL;qc"#FLLXY &qc* ++,s=DD<2D77D<