Seed Audio 1.0 — Text to Audio | NetRoom
ByteDance Sound and scene generation

Seed Audio 1.0

Describe a scene. Voice, ambience, music and effects come out in one generation.

Example output

Listen to Seed Audio 1.0

Original prompt

A calm, intimate English narrator says clearly: "After the rain, the garden wakes. Every small detail becomes a new story." Gentle warm delivery, soft distant rain ambience, natural pauses, no music.

Scene brief · Example request

A rainy night street. A woman says quietly: "We should go." Distant thunder, a car passing, soft piano underneath.

Up to 2,048 characters per brief.

Voice controls adjustable range
  • Pitch
    -1212
  • Speed
    -50100
  • Volume
    -50100

Voice controls: pitch -12 to 12, speed -50 to 100, volume -50 to 100.

In one generation

  • Voice: "We should go."
  • Ambience: rainy night street, distant thunder
  • Music: soft piano underneath
  • Effects: a car passing
$0.19
per minute of audio
Generate a scene
MP3 · WAV · FLAC · OGGFormats
2,048Scene description, characters
Pitch · Speed · VolumeVoice controls
Price

Pay per minute of audio

$0.19
per minute of audio

Billed by the minute of audio you generate.

Strengths

What it's the best tool for

  • Audio output from a written instruction
  • Speech pitch, speed and volume controls
  • Additional audio and image references
  • Export in MP3, WAV, FLAC and OGG
Limitations

When to reach for something else

  • Complex scenes need a detailed description
  • Output is a single file without separate stems
  • Length of one generation is limited
Where teams use it

Four scenarios where it pays for itself

01
Video sound design
Atmosphere and effects for clips
02
Games
Location ambiences from a brief
03
Podcasts
Intros and audio stingers
04
Advertising
A ready sound bed for a spot
About model

More about Seed Audio 1.0

Seed Audio 1.0: from a written direction to an audio track

Seed Audio 1.0 creates audio from a text instruction. Start with a small, clear passage: a short spoken line, narration, or a sound direction for a scene. In NetRoom, you can audition the output and download an available audio format. A short first test makes the voice character and instruction following easier to evaluate.

Separate the spoken words from the direction

For a narrator, write the actual sentence explicitly and describe the delivery separately: language, mood, pauses and intensity. Avoid conflicting instructions in one request. For a sound scene, identify the main events and their order. Audio and image references are available as additional inputs; their usefulness depends on the task and should be checked in the resulting audio.

What the example lets you hear

The page includes an actual model result from a short English prompt about a garden after rain. The prompt appears beside the track so you can compare the request with the delivery, pauses and background sound. The Russian prompt translation explains the intended scene; it does not mean that the English audio was generated in Russian.

Prepare the output for use

Listen to the complete file and check pronunciation, phrase boundaries, level and unwanted sounds. Pitch, speed and volume controls can adjust the delivery, but do not guarantee an exact acting performance. Available formats and the customer quote appear in NetRoom. Begin with a short instruction, download the result and decide which details to refine in another version.

Use Seed Audio 1.0 via the API

The same engine, straight from your code: one key and one balance for text, images, video and sound. Pay only for the requests you make.

curl
curl https://netroom.ai/api/v1/generations \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "bytedance/seed-audio-10", "input": {"prompt": "A rainy night street, distant thunder, a car passing, soft piano underneath"}}'

The model id is already in the example. The full parameter reference and prices live in GET /api/v1/models and in the docs.

API documentation Get an API key

Describe the scene. Get the sound.

You pay for the audio you generate, by the minute.