b6i:$ddlZddlZddlZddlZddlmZddlmZddlmZ ddl m Z ddl mZGd d eeZGd d eeZGd dej$dZGddej$dZGddZy#e $r e dwxYw)N)Enum)overload)Required)connectzgThe websockets package is required for realtime speech-to-text. Install it with: pip install websockets)RealtimeConnectionc,eZdZdZdZdZdZdZdZdZ dZ y ) AudioFormatz/Audio format options for realtime transcriptionpcm_8000 pcm_16000 pcm_22050 pcm_24000 pcm_44100 pcm_48000 ulaw_8000N) __name__ __module__ __qualname____doc__PCM_8000 PCM_16000 PCM_22050 PCM_24000 PCM_44100 PCM_48000 ULAW_8000e/root/.openclaw/workspace/visionaryfx/venv/lib/python3.12/site-packages/elevenlabs/realtime/scribe.pyr r s(9HIIIIIIrr ceZdZdZdZdZy)CommitStrategyz Strategy for committing transcription results. VAD: Voice Activity Detection - automatically commits when speech ends MANUAL: Manual commit - requires calling commit() to commit the segment vadmanualN)rrrrVADMANUALrrrr!r! s C Frr!ceZdZUdZeeed<eeed<eeed<e ed<e ed<e ed<eed<eed <eed <e ed <y ) RealtimeAudioOptionsa0 Options for providing audio chunks manually. Attributes: model_id: The model ID to use for transcription (required) audio_format: The audio format (required) sample_rate: The sample rate in Hz (required) commit_strategy: Strategy for committing transcriptions (optional, defaults to MANUAL) vad_silence_threshold_secs: Silence threshold in seconds for VAD (must be between 0.3 and 3.0) vad_threshold: Threshold for voice activity detection (must be between 0.1 and 0.9) min_speech_duration_ms: Minimum speech duration in milliseconds (must be between 50 and 2000) min_silence_duration_ms: Minimum silence duration in milliseconds (must be between 50 and 2000) language_code: An ISO-639-1 or ISO-639-3 language_code corresponding to the language of the audio file. Can sometimes improve transcription performance if known beforehand. include_timestamps: Whether to receive the committed_transcript_with_timestamps event after committing the segment (optional, defaults to False) model_id audio_format sample_ratecommit_strategyvad_silence_threshold_secs vad_thresholdmin_speech_duration_msmin_silence_duration_ms language_codeinclude_timestampsN) rrrrrstr__annotations__r intr!floatboolrrrr'r'+sRsm;''### %%  rr'F)totalcxeZdZUdZeeed<eeed<eed<eed<eed<e ed<e ed<eed <e ed <y ) RealtimeUrlOptionsa Options for streaming audio from a URL. Attributes: model_id: The model ID to use for transcription (required) url: The URL of the audio stream (required) commit_strategy: Strategy for committing transcriptions (optional, defaults to MANUAL) vad_silence_threshold_secs: Silence threshold in seconds for VAD (must be between 0.3 and 3.0) vad_threshold: Threshold for voice activity detection (must be between 0.1 and 0.9) min_speech_duration_ms: Minimum speech duration in milliseconds (must be between 50 and 2000) min_silence_duration_ms: Minimum silence duration in milliseconds (must be between 50 and 2000) language_code: An ISO-639-1 or ISO-639-3 language_code corresponding to the language of the audio file. Can sometimes improve transcription performance if known beforehand. include_timestamps: Whether to receive the committed_transcript_with_timestamps event after committing the segment (optional, defaults to False) r(urlr+r,r-r.r/r0r1N) rrrrrr2r3r!r5r4r6rrrr9r9GsG sm ### %%  rr9ceZdZdZddedefdZededefdZ ede defdZ de jee fdefd Z dedefd Z de defd Zd edd fdZ ddededede j ede j ede j ede j ede j ede j edefdZy )ScribeRealtimea+ Helper class for creating realtime speech-to-text connections. Supports two modes: 1. URL-based: Stream audio from a URL (uses ffmpeg for conversion) 2. Manual: Send audio chunks yourself Example (URL-based): ```python connection = await elevenlabs.speech_to_text.realtime.connect({ "model_id": "scribe_v2_realtime", "url": "https://stream.example.com/audio.mp3" }) ``` Example (Manual chunks): ```python connection = await elevenlabs.speech_to_text.realtime.connect({ "model_id": "scribe_v2_realtime", "audio_format": AudioFormat.PCM_16000, "sample_rate": 16000 }) # Send audio chunks await connection.send({"audio_base_64": chunk}) ``` api_keybase_urlc ||_||_yN)r=r>)selfr=r>s r__init__zScribeRealtime.__init__~s   roptionsreturnc Kywr@rrArCs rrzScribeRealtime.connect !c Kywr@rrFs rrzScribeRealtime.connectrGrHcKd|v}d|vr td|r1|jtjt|d{S|j tjt |d{S757w)a Create a realtime transcription connection. Args: options: Either RealtimeAudioOptions for manual chunk sending or RealtimeUrlOptions for URL streaming Returns: RealtimeConnection instance ready to send/receive data Raises: ValueError: If invalid options are provided RuntimeError: If ffmpeg is not available (for URL-based streaming) Example: ```python # URL-based streaming connection = await elevenlabs.speech_to_text.realtime.connect({ "model_id": "scribe_v2_realtime", "url": "https://stream.example.com/audio.mp3", }) # Manual chunks connection = await elevenlabs.speech_to_text.realtime.connect({ "model_id": "scribe_v2_realtime", "audio_format": AudioFormat.PCM_16000, "sample_rate": 16000, "commit_strategy": CommitStrategy.MANUAL }) ``` r:r(z/model_id is required for realtime transcriptionN) ValueError _connect_urltypingcastr9_connect_audior')rArC is_url_modes rrzScribeRealtime.connectstFw& W $NO O **6;;7I7+STT T,,V[[9Mw-WXX XUXs$AA>A:0A>5A<6A><A>c K|d}|jd}|jd}|jdtj}|jd}|jd}|jd}|jd} |jd } |jd d } |r|s td |j ||j |j |||| | | } t | d|jid{} t| |d}tj|j|_ |jd|S7Mw)z'Connect with manual audio chunk sendingr(r)r*r+r,r-r.r/r0r1Fz?audio_format and sample_rate are required for manual audio mode r(r)r+r,r-r.r/r0r1 xi-api-keyadditional_headersN websocketcurrent_sample_rateffmpeg_processopen)getr!r%rK_build_websocket_urlvaluewebsocket_connectr=rasyncio create_task_start_message_handler _message_task_emit)rArCr(r)r*r+r,r-r.r/r0r1ws_urlrW connections rrOzScribeRealtime._connect_audios\:&{{>2 kk-0 !++&79N9NO%,[[1M%N" O4 !(-E!F")++.G"H O4 $[[)=uE;^_ _**%+++11'A'#9$;'1+  ,  ,dll;  ( + $+#6#6z7X7X7Z#[   ! sDE E AEc K|d}|jd}|jdtj}|jd}|jd}|jd}|jd}|jd} |jd d } |s td d } tj } |j || j|j||||| | } t| d|jid{} tjdd|dddt| dddg tjtjd}t%|| |}t'j(|j+|_|j/dt'j(|j1||S7#t$r'|j!ddd{7t#dwxYww) z3Connect with URL-based audio streaming using ffmpegr(r:r+r,r-r.r/r0r1Fzurl is required for URL modei>rRrSrTNffmpegz-iz-fs16lez-arz-ac1-r)stdoutstderrbufsizeizffmpeg not foundziffmpeg is required for URL-based audio streaming. Please install ffmpeg: https://ffmpeg.org/download.htmlrVrZ)r[r!r%rKr rr\r]r^r= subprocessPopenr2PIPEDEVNULLFileNotFoundErrorclose RuntimeErrorrr_r`rarbrc_stream_ffmpeg_to_websocket)rArCr(r:r+r,r-r.r/r0r1r*r)rdrWrYres rrLzScribeRealtime._connect_urls:&kk% !++&79N9NO%,[[1M%N" O4 !(-E!F")++.G"H O4 $[[)=uE;< < ",, **%+++11'A'#9$;'1+  ,  ,dll;   '--#'3{+3 "!)) N*( +) $+#6#6z7X7X7Z#[    D<replaceappendr2lowerjoin) rAr(r)r+r,r-r.r/r0r1baseparams query_strings rr\z#ScribeRealtime._build_websocket_urles!}}$$Z:BB9gVz "L> */ 0  & 1 MM78R7ST U  $ MMN=/: ; ! - MM34J3KL M " . MM45L4MN O  $ MMN=/: ;  ) MM/4F0G0M0M0O/PQ Rxx' 3L>BBr)zwss://api.elevenlabs.io)NNNNNN)rrrrr2rBrr'rrr9rMUnionrOrLruOptionalr5r4r6r\rrrr<r<as8!!s!!%! !! !#! !! +Y24FFG+Y +YZ.,@.EW.`J*<JASJX,5B047;8<.248&C&C&C &C %+OOE$: &C u- &C!' 4&C"(!5&Cs+&C#OOD1&C &Crr<)r_r}rnrMenumrrtyping_extensionsrwebsockets.asyncio.clientrr^ ImportErrorrerr2r r! TypedDictr'r9r<rrrrs  &F+#tS$6++58))4jCjCk  2 s BB