Patch parameters

Translated

OpenAI	ElevenLabs	Teen V1 behavior
`input`	`text`	Speech text
`model`	`model_id`	Selects routing, then becomes `teen-v1`
`voice`	`voice_id`	Uses a TeenageWorks ID or provider voice fallback
`response_format`	`output_format`	Maps to PCM, MP3, or WAV
`cfg_value`	`cfg_value`	Passes through unchanged
`instructions` / `voice_description`	`voice_description`	Request-time voice description
`reference_audio`	`reference_audio`	Inline reference recording
`reference_audio_format`	`reference_audio_format`	Reference recording format
`reference_transcript` / `reference_text`	Same custom names	Canonical transcript or legacy alias
—	`seed`	Passes through unchanged
`timeout`	`request_options.timeout_in_seconds`	Sets the TeenageWorks request timeout

The patch adds the required language value internally.

Ignored with a warning

OpenAI instructions maps to voice_description; speed is ignored.

ElevenLabs enable_logging, optimize_streaming_latency, language_code, voice_settings, pronunciation_dictionary_locators, previous_text, next_text, previous_request_ids, next_request_ids, use_pvc_as_ivc, apply_text_normalization, and apply_language_text_normalization are ignored.

Each ignored value emits TeenageWorksUnsupportedParameterWarning. Provider voice fallback emits TeenageWorksVoiceMappingWarning.

Rejected

Unsupported codecs raise UnsupportedAudioFormatError.
Timestamp methods and temperature raise UnsupportedFeatureError.
Unknown arguments raise TypeError.

Nothing is silently discarded.

When inline voice-modelling fields are present, the provider voice ID is not used. voice_description and either transcript name cannot be combined.