ByteDanceSeedAudio - ComfyUI Built-in Node Documentation

Generate speech, music, sound effects and multi-speaker dialogue from a single prompt with ByteDance Seed Audio 1.0. Describe the voice(s), emotion, ambience, background music and sound effects in the prompt, and include the lines to speak. Optionally pick a built-in preset voice, clone voices from up to 3 reference clips (tagged @Audio1-3 in the prompt), or derive a voice from a character image. Up to 2 minutes of audio per run. The multilingual model supports 20 languages and timestamp-based timing control.

Inputs

Parameter	Description	Data Type	Required	Range
`text_prompt`	Describe the voice(s), emotion, pacing, ambience, background music and sound effects, and include the lines to speak (name characters inline for dialogue). In ‘audio reference’ mode, refer to connected clips by order as @Audio1, @Audio2, @Audio3. Maximum 3000 characters.	STRING	Yes	1 to 3000 characters
`reference_mode`	How to condition the voice: ‘text only’ (describe everything in the prompt), ‘audio reference’ (clone up to 3 voices, tagged @Audio1-3), ‘image reference’ (derive a voice from one character image), or ‘preset voice’ (pick a built-in named voice that reads the prompt).	COMBO	Yes	`"text only"` `"audio reference"` `"image reference"` `"preset voice"`
`reference_audio_1`	Reference clip for voice cloning, tagged @Audio1 in the prompt. Up to 30s. Only available when `reference_mode` is “audio reference”.	AUDIO	No	Up to 30 seconds
`reference_audio_2`	Reference clip tagged @Audio2 in the prompt. Up to 30s. Only available when `reference_mode` is “audio reference”.	AUDIO	No	Up to 30 seconds
`reference_audio_3`	Reference clip tagged @Audio3 in the prompt. Up to 30s. Only available when `reference_mode` is “audio reference”.	AUDIO	No	Up to 30 seconds
`reference_image`	A single character image; the model derives a voice from it. Cannot be combined with reference audio. Only available when `reference_mode` is “image reference”.	IMAGE	No	-
`preset_voice`	A built-in TTS 2.0 voice that reads the prompt. No reference clip needed, and @AudioN tags are not used in this mode. Only available when `reference_mode` is “preset voice”.	COMBO	No	Multiple options available (see description)
`sample_rate`	Output sample rate in Hz. (default: “24000”)	COMBO	Yes	`"8000"` `"16000"` `"24000"` `"32000"` `"44100"` `"48000"`
`speech_rate`	Speaking speed. 0 = normal, 100 = 2.0x, -50 = 0.5x. (default: 0)	INT	Yes	-50 to 100
`loudness_rate`	Loudness. 0 = normal, 100 = 2.0x, -50 = 0.5x. (default: 0)	INT	Yes	-50 to 100
`pitch_rate`	Pitch shift in semitones (-12 to 12). (default: 0)	INT	Yes	-12 to 12
`seed`	Seed controls whether the node should re-run; results are non-deterministic regardless of seed. (default: 42)	INT	Yes	0 to 2147483647
`model`	Model version. The multilingual variant supports 20 languages and per-sentence timing via `[5.5s:8.0s]` timestamps. (default: “seed-audio-1.0-multilingual”)	COMBO	No	`"seed-audio-1.0-multilingual"` `"seed-audio-1.0"`

Parameter Constraints

Reference mode dependencies: The reference_mode parameter determines which other inputs are required:
- “text only”: No additional inputs required. The prompt must not contain @AudioN tags.
- “audio reference”: Requires at least one of reference_audio_1, reference_audio_2, or reference_audio_3 to be connected. Reference clips must be connected in order without gaps (e.g., _1, then _2, then _3). Each clip is limited to 30 seconds maximum duration. The prompt must reference connected clips using @Audio1, @Audio2, @Audio3 tags.
- “image reference”: Requires reference_image to be connected. The prompt must not contain @AudioN tags.
- “preset voice”: Requires preset_voice to be selected. The prompt must not contain @AudioN tags (the entire prompt is read in the selected voice).
Audio reference ordering: When using “audio reference” mode, reference audio inputs must be connected sequentially starting from reference_audio_1 without gaps. For example, you can connect _1 and _2, but not _1 and _3 without _2.
Maximum audio tags: The prompt can reference up to 3 audio clips (@Audio1, @Audio2, @Audio3) when in “audio reference” mode. The highest numbered tag must not exceed the number of connected reference audio inputs.
Model differences: The “seed-audio-1.0-multilingual” model supports 20 languages (English, Chinese, Japanese, Korean, Mexican & Castilian Spanish, Indonesian, German, Brazilian Portuguese, French, Thai, Vietnamese, Malay, Filipino, Italian, Russian, Dutch, Polish, Turkish, Swedish) plus per-sentence timing control using timestamps in the format [5.5s:8.0s]. The “seed-audio-1.0” model supports English and Chinese only, without timing control.

Outputs

Output Name	Description	Data Type
`AUDIO`	The generated audio output from ByteDance Seed Audio 1.0, containing speech, music, sound effects, or multi-speaker dialogue as described in the prompt.	AUDIO

This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHub

Source fingerprint (SHA-256): e86e4edde424b4427d864350a9d3b082e271fbd2b1e335175637a9cc3ad51163

​Inputs

​Parameter Constraints

​Outputs

Inputs

Parameter Constraints

Outputs