This document provides a comprehensive reference for the parameters available across various audio generation models in the Scenario API. Each model has a unique modelId and a set of specific parameters that can be used to control the speech generation process. Understanding these parameters is crucial for effectively utilizing the API to achieve desired audio outputs.

Below, you will find detailed information for each audio model, including its modelId, the types of parameters it accepts, allowed values, default settings, and a clear description of each parameter's function.

ElevenLabs

ElevenLabs V3

Model ID: model_elevenlabs-tts-v3

Input	Label	Type	Default	Min	Max	Allowed Values	Notes
`text`	Text	string	–	–	–	–	Required. Up to 40k characters
`voice`	Voice	select	Aria	–	–	"Aria", "Roger", "Sarah", "Laura", "Charlie", "George", "Callum", "River", "Liam", "Charlotte", "Alice", "Matilda", "Will", "Jessica", "Eric", "Chris", "Brian", "Daniel", "Lily", "Bill"
`stability`	Stability	number	0.5	0	1	–
`similarityBoost`	Similarity Boost	number	0.5	0	1	–
`style`	Style Exaggeration	number	0	0	1	–
`speed`	Speed	number	1	0.7	1.2	–	<1 slows; >1 speeds up
`previousText`	Previous Text	string	–	–	–	–	optional context
`nextText`	Next Text	string	–	–	–	–	optional context
`languageCode`	Language Code	select	""	–	–	ISO 639‑1 codes

ElevenLabs Turbo v2.5

Model ID: elevenlabs-turbo-v2-5

Input	Label	Type	Default	Min	Max	Allowed Values	Notes
`text`	Text	`string`	–	–	–	–	Required. Text to convert to speech (max 40000 chars)
`voice`	Voice	`string`	Aria	–	–	Aria, Roger, Sarah, Laura, Charlie, George, Callum, River, Liam, Charlotte, Alice, Matilda, Will, Jessica, Eric, Chris, Brian, Daniel, Lily, Bill	Voice preset
`stability`	Stability	`number`	0.5	0	1	–	Controls voice stability
`similarityBoost`	Similarity Boost	`number`	0.5	0	1	–	Closeness to selected voice
`styleExaggeration`	Style Exaggeration	`number`	0	0	1	–	Boosts emotional expression
`speed`	Speed	`number`	1	0.7	1.2	–	<1 slows, >1 speeds up
`previousText`	Previous Text	`string`	–	–	–	–	Optional. Helps continuity across multi-part generation (max 10000 chars)
`nextText`	Next Text	`string`	–	–	–	–	Optional. Helps continuity (max 10000 chars)
`languageCode`	Language Code	`string`	""	–	–	"" (auto), en, ca, es, fr, de, it, ja, ko, zh, ru, ar, hi, bn, pa, ta, te, mr, ur, fa, tr, nl, sv, da, no, fi, el, ro, hu, cs, sk, sl, pt, id, th, vi, ms, tl, yo, ig, ha, am, az, be, bg, hr	Forces language for synthesis

ElevenLabs Multilingual v2

Model ID: model_elevenlabs-multilingual-v2

Input	Label	Type	Default	Min	Max	Allowed Values	Notes
`text`	Text	string	–	–	–	–	Required. Up to 40k characters
`voice`	Voice	select	Aria	–	–	"Aria", "Roger", "Sarah", "Laura", "Charlie", "George", "Callum", "River", "Liam", "Charlotte", "Alice", "Matilda", "Will", "Jessica", "Eric", "Chris", "Brian", "Daniel", "Lily", "Bill"
`stability`	Stability	number	0.5	0	1	–
`similarityBoost`	Similarity Boost	number	0.5	0	1	–
`style`	Style Exaggeration	number	0	0	1	–
`speed`	Speed	number	1	0.7	1.2	–	<1 slows; >1 speeds up
`previousText`	Previous Text	string	–	–	–	–	optional context
`nextText`	Next Text	string	–	–	–	–	optional context
`languageCode`	Language Code	select	""	–	–	ISO 639‑1 codes

Minimax

Minimax Speech 2.6 HD

Model ID: model_minimax-speech-2-6-hd

Input	Label	Type	Default	Min	Max	Allowed Values
`text`	Text	string	–	–	–	–
`voiceId`	Voice Id	select	Wise_Woman	–	–	Wise_Woman, Friendly_Person, Inspirational_girl, Deep_Voice_Man, Calm_Woman, Casual_Guy, Lively_Girl, Patient_Man, Young_Knight, Determined_Man, Lovely_Girl, Decent_Boy, Imposing_Manner, Elegant_Man, Abbess, Sweet_Girl_2, Exuberant_Girl
`speed`	Speed	number	1	0.5	2	–
`volume`	Volume	number	1	0	10	–
`pitch`	Pitch	number	0	-12	12	–
`emotion`	Emotion	select	auto	–	–	auto, neutral, happy, sad, angry, fearful, disgusted, surprised
`englishNormalization`	English Normalization	boolean	false	–	–	–
`sampleRate`	Sample Rate	number	32000	–	–	8000, 16000, 22050, 24000, 32000, 44100
`bitrate`	Bitrate	number	128000	–	–	32000, 64000, 128000, 256000
`channel`	Channel	select	mono	–	–	mono, stereo
`languageBoost`	Language Boost	select	Automatic	–	–	(list of 25 language options)

Minimax Speech 2.6 Turbo

Model ID: model_minimax-speech-2-6-turbo

Input	Label	Type	Default	Min	Max	Allowed Values
`text`	Text	string	–	–	–	–
`voiceId`	Voice Id	select	Wise_Woman	–	–	Wise_Woman, Friendly_Person, Inspirational_girl, Deep_Voice_Man, Calm_Woman, Casual_Guy, Lively_Girl, Patient_Man, Young_Knight, Determined_Man, Lovely_Girl, Decent_Boy, Imposing_Manner, Elegant_Man, Abbess, Sweet_Girl_2, Exuberant_Girl
`speed`	Speed	number	1	0.5	2	–
`volume`	Volume	number	1	0	10	–
`pitch`	Pitch	number	0	-12	12	–
`emotion`	Emotion	select	auto	–	–	auto, neutral, happy, sad, angry, fearful, disgusted, surprised
`englishNormalization`	English Normalization	boolean	false	–	–	–
`sampleRate`	Sample Rate	number	32000	–	–	8000, 16000, 22050, 24000, 32000, 44100
`bitrate`	Bitrate	number	128000	–	–	32000, 64000, 128000, 256000
`channel`	Channel	select	mono	–	–	mono, stereo
`languageBoost`	Language Boost	select	Automatic	–	–	(list of 25 language options)

Tada

Tada 1B Text to Speech

Model ID: model_tada-1b-text-to-speech

Input	Label	Type	Default	Min	Max	Allowed Values	Notes
`audio`	Reference Audio	`file`	None	None	None	None	Required. Reference audio for voice cloning.
`prompt`	Prompt	`string`	None	None	10000	None	Required. Text to synthesize with the reference voice.
`transcript`	Reference Transcript	`string`	""	None	None	None	Transcript of the reference audio. Required for non-English references.
`language`	Language	`string`	en	None	None	en, ar, ch, de, es, fr, it, ja, pl, pt	Language used for text alignment.
`outputFormat`	Output Format	`string`	wav	None	None	wav, mp3	Output audio file format.
`numInferenceSteps`	Inference Steps	`number`	20	1	50	None	Number of ODE solver steps for acoustic generation.
`speedUpFactor`	Speed Up Factor	`number`	1	0.5	2	None	Values > 1 speed up and values < 1 slow down speech.
`temperature`	Temperature	`number`	0.6	0	2	None	Sampling temperature for text token generation.
`topP`	Top P	`number`	0.9	0	1	None	Top-p nucleus sampling value.
`repetitionPenalty`	Repetition Penalty	`number`	1.1	1	2	None	Penalty applied to repeated tokens.
`acousticCfgScale`	Acoustic CFG Scale	`number`	1.6	0	10	None	Classifier-free guidance scale for acoustic generation.
`noiseTemperature`	Noise Temperature	`number`	0.9	0	2	None	Temperature for diffusion noise during flow matching.
`numExtraSteps`	Extra Steps	`number`	0	0	50	None	Additional autoregressive steps for continuation.

Tada 3B Text to Speech

Model ID: model_tada-3b-text-to-speech

Input	Label	Type	Default	Min	Max	Allowed Values	Notes
`audio`	Reference Audio	`file`	None	None	None	None	Required. Reference audio for voice cloning.
`prompt`	Prompt	`string`	None	None	10000	None	Required. Text to synthesize with the reference voice.
`transcript`	Reference Transcript	`string`	""	None	None	None	Transcript of the reference audio. Required for non-English references.
`language`	Language	`string`	en	None	None	en, ar, ch, de, es, fr, it, ja, pl, pt	Language used for text alignment.
`outputFormat`	Output Format	`string`	wav	None	None	wav, mp3	Output audio file format.
`numInferenceSteps`	Inference Steps	`number`	20	1	50	None	Number of ODE solver steps for acoustic generation.
`speedUpFactor`	Speed Up Factor	`number`	1	0.5	2	None	Values > 1 speed up and values < 1 slow down speech.
`temperature`	Temperature	`number`	0.6	0	2	None	Sampling temperature for text token generation.
`topP`	Top P	`number`	0.9	0	1	None	Top-p nucleus sampling value.
`repetitionPenalty`	Repetition Penalty	`number`	1.1	1	2	None	Penalty applied to repeated tokens.
`acousticCfgScale`	Acoustic CFG Scale	`number`	1.6	0	10	None	Classifier-free guidance scale for acoustic generation.
`noiseTemperature`	Noise Temperature	`number`	0.9	0	2	None	Temperature for diffusion noise during flow matching.
`numExtraSteps`	Extra Steps	`number`	0	0	50	None	Additional autoregressive steps for continuation.

Lux TTS

Model ID: model_lux-tts

Input	Label	Type	Default	Min	Max	Allowed Values	Notes
`prompt`	Prompt	`string`	None	None	10000	None	Required. Text to convert to speech.
`audio`	Reference Audio	`file`	None	None	None	None	Required. Reference audio for voice cloning.
`guidanceScale`	Guidance Scale	`number`	3	0	10	None	Higher values increase adherence to the reference voice.
`numInferenceSteps`	Inference Steps	`number`	4	1	16	None	Number of flow-matching inference steps.
`maxRefLength`	Max Reference Length	`number`	5	1	15	None	Maximum reference audio duration used for voice encoding (seconds).
`seed`	Seed	`number`	None	0	2147483647	None	Seed for reproducible outputs.