ElevenLabs patch

1. Install and configure

Install the elevenlabs extra, then set both TEENAGEWORKS_API_KEY and the ElevenLabs credential used by your application. The provider credential remains necessary for calls that pass through.

python -m pip install "teenageworks[elevenlabs]"

2. Patch before creating the client

from teenageworks import patch_elevenlabs

patch_elevenlabs()

3. Keep the existing ElevenLabs call

from elevenlabs.client import ElevenLabs

client = ElevenLabs()
chunks = client.text_to_speech.convert(
    voice_id="provider-voice-id",
    text="This call is routed to Teen V1.",
    model_id="eleven_multilingual_v2",
    voice_description="Young, warm, conversational narrator",
    output_format="mp3_44100_128",
)

The patched method accepts custom request-time voice_description and reference fields. When one is supplied, the provider voice ID is not used for the TeenageWorks request.

teen-v1, models beginning with eleven_, and the omitted provider default route to TeenageWorks. Unknown custom model IDs pass through to ElevenLabs.

The patch covers tested synchronous and asynchronous convert() and stream() methods, supported persistent text-input sockets, and raw-response variants.