Seed Audio 1.0 is BytePlus's speech-generation model for converting text into downloadable spoken audio. Through PiAPI, you can generate speech, guide delivery with permitted reference audio or one image, and control rate, pitch, loudness, format, and sample rate.
PiAPI Signal Studio
Connected playgroundGenerate Seed Audio 1.0 speech, upload permitted references, poll the task, and copy the matching API request.
Billing is based on successful output duration at $0.002 per second. A 120-second maximum output would cost about $0.24 before any account-specific rules.
Live result
Ready to generate audio
Enter a script, choose optional references and settings, then generate audio through PiAPI.
Format
WAV
Sample rate
24 kHz
References
Default
Duration
≤120 sec
Matching API request
{
"model": "byteaudio",
"task_type": "seed-audio-1.0",
"input": {
"text": "Welcome to PiAPI Signal Studio. This Seed Audio 1.0 test turns a short product script into clear spoken audio for an app demo, narration, or video voiceover.",
"format": "wav",
"sample_rate": 24000,
"speech_rate": 0,
"pitch_rate": 0,
"loudness_rate": 0
}
}Generated audio
Listen to a result generated through Signal Studio, with its script, mode, settings, and output duration included.
Welcome to PiAPI Signal Studio. This Seed Audio 1.0 test turns a short product script into clear spoken audio for an app demo, narration, or video voiceover.
Model overview
Seed Audio 1.0 is a BytePlus speech-generation model available through PiAPI as an asynchronous text-to-speech API.
PiAPI accepts a script of up to 2,048 characters, sends the generation task for processing, and returns a temporary audio URL when the task completes.
The integration supports standard speech generation and optional voice or style guidance. You can use permitted reference audio or one reference image, then control speech rate, pitch, loudness, format, and sample rate.
PiAPI's seed-audio-1.0 task focuses on spoken audio. Use a dedicated music or video-sound model when the project requires songs, ambience, or sound effects.
API facts checked July 9, 2026 against the PiAPI Seed Audio 1.0 documentation.
Core capabilities
Seed Audio 1.0 through PiAPI is built for spoken audio workflows: generate voice from text, guide delivery with short reference clips, and assign different voices inside one script.
Generate spoken audio from text with BytePlus seed-audio-1.0 through PiAPI's asynchronous task API.
Use permitted reference audio for voice cloning or voice/style guidance. PiAPI accepts up to three audio references and recommends 5-10 seconds of clean reference audio.
Assign up to three reference voices in one script with @Audio1, @Audio2, and @Audio3 markers at natural word or sentence boundaries.
Voice direction
Move beyond a basic text-to-speech request with permitted reference audio, multiple speaker markers, or one reference image.
Upload WAV, MP3, PCM, or OGG Opus audio up to 10 MB. Public URL references must be 30 seconds or shorter.
PiAPI recommends 5–10 seconds of clean reference audio.
Add up to three references and use @Audio1, @Audio2, or @Audio3 to choose the voice for a dialogue turn.
Keep speaker changes at natural word or sentence boundaries.
Use one JPEG, PNG, or WebP image up to 10 MB. Audio and image references cannot be mixed in one request.
Choose image guidance instead of audio references for the request.
Upload or reference a voice only when you have the speaker's permission or another lawful basis to use it. Do not use the model for deceptive impersonation or false endorsement.
Audio settings
Start with the defaults or adjust the voice delivery and output requirements before creating a task.
Developer quick start
Create a task, poll its status, and download the audio URL returned after completion.
Send the script and settings to PiAPI's standard task endpoint.
Check the task ID until its status becomes completed or failed.
Use output.audio_url promptly because the result URL is temporary.
curl --request POST \
--url https://api.piapi.ai/api/v1/task \
--header "Content-Type: application/json" \
--header "X-API-Key: YOUR_API_KEY" \
--data '{
"model": "byteaudio",
"task_type": "seed-audio-1.0",
"input": {
"text": "Hello from PiAPI. This is a Seed Audio 1.0 test.",
"format": "mp3",
"sample_rate": 24000
}
}'Usage pricing
Pay for successfully generated output duration instead of a subscription-style model plan.
Output audio
$0.002
per second
Approximately $0.12 per generated minute.
Production workflows
Use the PiAPI endpoint for spoken-word workflows that benefit from repeatable scripts, voice guidance, and controllable delivery.
Turn product demos, tutorials, advertisements, and social scripts into downloadable voice tracks.
Generate prepared spoken segments and reuse permitted reference voices across related content.
Tune rate, pitch, and loudness for concise product messages, onboarding, and training material.
Assign up to three permitted reference voices with @Audio markers for short scripted conversations.
Clear boundaries
Review the technical and consent boundaries before building a production workflow.
Common questions
Seed Audio 1.0 is a BytePlus speech-generation model available through PiAPI. It converts text into downloadable audio and supports permitted reference audio, multiple voice markers, image guidance, and speech-output controls.
Yes. PiAPI exposes Seed Audio 1.0 through POST https://api.piapi.ai/api/v1/task using the byteaudio model and seed-audio-1.0 task type. The task is asynchronous, so applications poll the returned task ID until generation completes or fails.
PiAPI lists Seed Audio 1.0 at $0.002 per successful output second, approximately $0.12 per minute. Output duration is rounded up to the nearest second. Failed validation, moderation, and upstream requests refund the credits frozen for the task.
The PiAPI endpoint accepts permitted reference audio for voice cloning or voice and style guidance. You can upload an audio file, provide Base64 audio, or use a publicly accessible HTTPS audio URL. Only use reference voices when you have permission.
One Seed Audio 1.0 request can include up to three audio references. Use @Audio1, @Audio2, and @Audio3 in the script to assign the corresponding reference voice.
Generated audio can use WAV, MP3, PCM, or OGG Opus. Supported sample rates are 8, 16, 24, 32, 44.1, and 48 kHz. WAV at 24 kHz is the default.
One request can return up to 120 seconds of audio. The script itself is limited to 2,048 Unicode characters.
No. PiAPI's current Seed Audio 1.0 task is asynchronous. Submit a task, poll its status, and download the returned audio after it completes.
The endpoint accepts one JPEG, PNG, or WebP reference image up to 10 MB. Image and audio references cannot be mixed in the same request.
Common causes include empty or prohibited text, an unsupported file, a reference larger than 10 MB, reference audio longer than 30 seconds, an invalid reference entry, or upstream moderation.
Commercial use depends on the applicable PiAPI, BytePlus, voice, content, and jurisdictional terms. Review the current terms and confirm that you have permission for every reference voice before publishing or monetizing generated audio.
Related models
Choose a model based on whether the workflow needs spoken voice, video-matched sound, music, or a downstream video result.
Explore an alternative reference-audio text-to-speech workflow.
ExploreGenerate audio designed to match the action and timing of a video.
ExploreGenerate music when the project needs a complete track rather than speech.
ExploreBring a generated speech track into an audio-first AI video workflow.
ExploreSeed Audio 1.0 through PiAPI
Open Signal Studio to test a script, copy the matching API request, or use the PiAPI documentation to prepare your integration.