o lWi:@sddlZddlZddlZddlZddlmZddlmZddlmZzddl m Z Wn e y5e dwddl mZGd d d eeZGd d d eeZGd ddejddZGdddejddZGdddZdS)N)Enum)overload)Required)connectzgThe websockets package is required for realtime speech-to-text. Install it with: pip install websockets)RealtimeConnectionc@s,eZdZdZdZdZdZdZdZdZ dZ d S) 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_8000rr^/var/www/html/asistente-voz-ia/venv/lib/python3.10/site-packages/elevenlabs/realtime/scribe.pyrsrc@seZdZdZdZdZdS)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)rrrrVADMANUALrrrrr src@sneZdZUdZeeed<eeed<eeed<e ed<e ed<e ed<eed<eed <eed <e ed <d S) 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__rintrfloatboolrrrrr"+s     r"F)totalc@sbeZdZUdZeeed<eeed<eed<eed<eed<e ed<e ed<eed <e ed <d S) 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*r+r,N) rrrrrr-r.rr0r/r1rrrrr3Gs    r3c@seZdZdZd dedefddZededefd d Z ede defd d Z de j ee fdefd d Z dedefd dZ de defddZdeddfddZ      d!dededede jede jede jede jede jede jedefddZdS)"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}) ``` wss://api.elevenlabs.ioapi_keybase_urlcCs||_||_dSN)r7r8)selfr7r8rrr__init__~s zScribeRealtime.__init__optionsreturncdSr9rr:r<rrrrzScribeRealtime.connectcr>r9rr?rrrrr@csNd|v}d|vr td|r|tt|IdHS|tt|IdHS)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 }) ``` r4r#z/model_id is required for realtime transcriptionN) ValueError _connect_urltypingcastr3_connect_audior")r:r< is_url_moderrrrs#c s|d}|d}|d}|dtj}|d}|d}|d}|d} |d } |d d } |r9|s=td |j||j|j|||| | | d } t| d|jidIdH} t| |dd}t | |_ | d|S)z'Connect with manual audio chunk sendingr#r$r%r&r'r(r)r*r+r,Fz?audio_format and sample_rate are required for manual audio mode r#r$r&r'r(r)r*r+r, xi-api-keyadditional_headersN websocketcurrent_sample_rateffmpeg_processopen)getrr!rA_build_websocket_urlvaluewebsocket_connectr7rasyncio create_task_start_message_handler _message_task_emit)r:r<r#r$r%r&r'r(r)r*r+r,ws_urlrL connectionrrrrEsH           zScribeRealtime._connect_audioc sD|d}|d}|dtj}|d}|d}|d}|d}|d} |d d } |s6td d } tj} |j|| |j||||| | d } t| d|j idIdH}zt j dd|dddt | dddg t j t jdd}Wnty|ddIdHtdwt|| |d}t||_|dt|||S) z3Connect with URL-based audio streaming using ffmpegr#r4r&r'r(r)r*r+r,Fzurl is required for URL modei>rGrHrINffmpegz-iz-fs16lez-arz-ac1-r)stdoutstderrbufsizeizffmpeg not foundziffmpeg is required for URL-based audio streaming. Please install ffmpeg: https://ffmpeg.org/download.htmlrKrO)rPrr!rArrrQrRrSr7 subprocessPopenr-PIPEDEVNULLFileNotFoundErrorclose RuntimeErrorrrTrUrVrWrX_stream_ffmpeg_to_websocket)r:r<r#r4r&r'r(r)r*r+r,r%r$rYrLrNrZrrrrBsr           zScribeRealtime._connect_urlrZNcs|jr|jjs dSzzTd}t} |d|jjj|IdH}|s#n;t|d}z | d|iIdHWnt yT}z| ddd|iWYd}~nd}~wwt dIdHqWnt y|}z| ddd |iWYd}~nd}~wwW|jr|j z |jjd d WdStjy|j YdSwdS|jr|j z |jjd d Wwtjy|j Ywww) z-Stream audio from ffmpeg process to WebSocketNi Tzutf-8 audio_base_64errorzFailed to send audio: g{Gz?zFFmpeg streaming error: r)timeout)rNr_rTget_event_looprun_in_executorreadbase64 b64encodedecodesend ExceptionrXsleepkillwaitrbTimeoutExpired)r:rZ chunk_sizeloopchunk chunk_base64errrri7sZ  "  z*ScribeRealtime._stream_ffmpeg_to_websocketr#r$r&r'r(r)r*r+r,c Cs|jdddd} d|d|d|g} |dur%| d ||dur1| d ||dur=| d ||durI| d ||durU| d || dura| d| d| } | d| S)z-Build the WebSocket URL with query parameterszhttps://zwss://zhttp://zws://z model_id=z audio_format=zcommit_strategy=Nzvad_silence_threshold_secs=zvad_threshold=zmin_speech_duration_ms=zmin_silence_duration_ms=zlanguage_code=zinclude_timestamps=&z/v1/speech-to-text/realtime?)r8replaceappendjoin) r:r#r$r&r'r(r)r*r+r,baseparams query_stringrrrrQes& z#ScribeRealtime._build_websocket_url)r6)NNNNNN)rrrrr-r;rr"rrr3rCUnionrErBriOptionalr0r/r1rQrrrrr5asd    -0L3   r5)rTrprbrCenumrrtyping_extensionsrwebsockets.asyncio.clientrrS ImportErrorrZrr-rr TypedDictr"r3r5rrrrs(