Parameter Guide

Every field available in this screen and what it changes.

General

Text to generate

The final phrase the cloned voice should say. Guests can use up to 100 characters; signed-in users can use up to 500. Example: The airdrop has been dispatched to Karichar!

Reference audio

Upload an audio file such as WAV, MP3, MP4/M4A, OGG, or FLAC. The server normalizes it to a temporary WAV before inference.

Reference clip

The trimmed part of the reference file used by the model. Time fields accept 4.35 or 4:35 as 4 minutes 35 seconds; use 4.35s for literal seconds.

Clip length seconds

Sets a fixed selection length that can be moved across the uploaded audio. Default is 30 seconds.

Lock duration

Keeps the selected clip duration unchanged while you move either the start or end handle.

MAX

Uses the full uploaded audio file (up to 1 minute) and locks manual clip timing controls.

Output filename

Base name for generated audio files and the downloaded ZIP.

Versions

How many different versions to generate with current settings.

Seed

Optional deterministic seed. Empty means random; multiple versions increment the seed.

Expression

Optional expression style for the generated line. Example: angry, intense, shouting loudly.

CFG

Classifier-free guidance strength. The service allows 1.0–5.0; values around 1–3 are the recommended range. Values of 4–5 are experimental and higher guidance can over-constrain the sound or introduce artifacts.

Timesteps

Inference steps, capped at 30. More steps can refine the result, but increase generation time and GPU use with diminishing returns.

Post process

Radio

Radio post-processing

Optional effect pass. Disabled means the ZIP will contain only clean generated WAV files.

Low / High Hz

Band-pass range for the radio voice effect.

Transition

Soft edge width around the radio band-pass filter.

Noise

Amount of filtered static mixed into the voice.

Drive

Saturation intensity for compressed radio-style distortion.

Crackle level / rate

Volume and frequency of short random radio clicks.

Squelch

Adds short noise bursts before and after the generated voice to immitate the sound of a radio squelch.

Peak

Final normalization target. Must stay between 0 and 1.

Radio seed

Deterministic seed used only for noise, crackle, and squelch randomness.

Quick tips

Use a clear 10–30 second sample with one speaker and as little background noise as possible.

The selected reference clip is what drives the clone. Preview it before generating.

When recording, allow microphone access and keep a steady distance from the microphone.

WAV, MP3, MP4/M4A, OGG, FLAC and WEBM are accepted. The server normalizes reference audio to mono WAV.

More versions and higher timesteps can increase generation time and credit usage.

Credits

Before a signed-in generation starts, the server estimates the usage and temporarily reserves those credits.

The reservation is not the final charge. After completion, the measured generation and optional post-processing time determine the final cost; unused reserved credits return to your balance.

Failed or interrupted jobs release their reservation. Your balance cannot become negative. Anonymous demo generations do not use an account balance and are limited to one version.

Simple Voice Clone - TTS

Generate voice variants from a reference clip, optionally add radio post-processing, and download a ZIP.

Calculating…

Intended use AI disclosure

Voice Prompt

Text, delivery style, and reference audio selection.

Text to generate: — Expression

Reference audio

Output filename Output format

Reference Clip

Select the exact part of the uploaded audio that should drive the voice clone.

Clip length seconds MAX

Start End

Total: no fileSelected: 0.00s

Available endpoints

GET /api/v1/billing/policy — read the active server-side billing policy.

GET /api/v1/account — read the API key owner's account balance.

POST /api/v1/generate — submit a generation; it requires the generation:write API key scope and is currently disabled.

GET /api/v1/jobs — list recent jobs owned by the API key.

GET /api/v1/jobs/{request_id} — read one job's status and billing details.

GET /api/v1/jobs/{request_id}/download — download a completed ZIP.

API keys for integrations

Create one API key from the signed-in Account panel. The raw secret is displayed once and is never stored or recoverable; if it is lost, create a replacement. Do not put an API key into browser source code or commit it to a repository. CLI and external integrations send the generated secret in X-API-Key on /api/v1 requests.

Generation request for API v1

Field	Type	Purpose
`sample`	Audio file	WAV, MP3, MP4/M4A, OGG, FLAC, or WEBM reference voice clip. The server normalizes it before inference.
`text`	string	Text to generate.
`intended_use`	string	Intended use, for example `Game`, `Fictional scene`, or `Satire/parody`.
`ai_disclosure_accepted`	boolean	Must be `true` after reviewing the AI disclosure.
`control`	string	Optional control/style prefix.
`prompt_text`	string	Optional transcript for Ultimate Cloning. Do not combine it with `control`: upstream VoxCPM ignores the style prefix in that mode.
`duration`	number	Reference audio duration in seconds.
`cfg`	number	Classifier-free guidance strength, limited to `1.0..5.0`; the recommended range is about `1..3`.
`output_format`	string	Output format: `wav`, `flac`, `ogg`, `mp3`, or `m4a`.
`count`	integer	Number of requested outputs, up to the configured maximum (default 10).
`timesteps`	integer	Inference timesteps, limited to 1..30; the model documentation recommends 4..30.
`client_request_id`	string	Unique ID for one logical generation. Reusing an existing ID returns `409 Conflict` because that job already exists; use `GET /api/v1/jobs/{request_id}` to inspect it and choose a new ID for a new generation.

Field

Type

Purpose

sample

Audio file

WAV, MP3, MP4/M4A, OGG, FLAC, or WEBM reference voice clip. The server normalizes it before inference.

text

string

Text to generate.

intended_use

string

Intended use, for example Game, Fictional scene, or Satire/parody.

ai_disclosure_accepted

boolean

Must be true after reviewing the AI disclosure.

control

string

Optional control/style prefix.

prompt_text

string

Optional transcript for Ultimate Cloning. Do not combine it with control: upstream VoxCPM ignores the style prefix in that mode.

duration

number

Reference audio duration in seconds.

cfg

number

Classifier-free guidance strength, limited to 1.0..5.0; the recommended range is about 1..3.

output_format

string

Output format: wav, flac, ogg, mp3, or m4a.

count

integer

Number of requested outputs, up to the configured maximum (default 10).

timesteps

integer

Inference timesteps, limited to 1..30; the model documentation recommends 4..30.

client_request_id

string

Unique ID for one logical generation. Reusing an existing ID returns 409 Conflict because that job already exists; use GET /api/v1/jobs/{request_id} to inspect it and choose a new ID for a new generation.

Post-processing parameters

Set apply_post=true to enable one post-processing profile. Each profile adds its own fields to the request.

Radio

Field	Type	Purpose
`radio_low_hz`	number	Radio band lower edge, 0..24000 Hz; default 300.
`radio_high_hz`	number	Radio band upper edge, 0..24000 Hz and above `radio_low_hz`; default 3000.
`radio_transition_hz`	number	Filter transition width, 0..24000 Hz; default 150.
`radio_noise_level`	number	Radio noise level, 0..1; default 0.08.
`radio_drive`	number	Radio saturation drive, 0.1..10; default 2.2.
`radio_crackle_level`	number	Crackle intensity, 0..1; default 0.35.
`radio_crackle_rate`	number	Crackle rate, 0..1000; default 8.
`output_peak`	number	Output normalization peak, 0.01..1; default 0.92.
`radio_seed`	integer	Deterministic radio randomness seed, -2147483648..2147483647; default 1337.
`radio_squelch`	boolean	Enable radio squelch bursts; default false.
`radio_squelch_ms`	number	Squelch duration, 0..10000 ms; default 80.
`radio_squelch_level`	number	Squelch level, 0..1; default 0.12.

Check a job and download its ZIP

POST /api/v1/generate returns 202 Accepted with a job_id. Poll the job endpoint until its status is ready, then download the result with the same API key. If the submitted client_request_id is already used, the API returns 409 Conflict and identifies the existing job; use a new ID for another generation.

$apiKey = "API_KEY" $jobId = "generation-002" # Check the current status curl.exe -H "X-API-Key: $apiKey" ` "https://voiclone.it/api/v1/jobs/$jobId" # Download after the response reports status=ready and download_available=true curl.exe -H "X-API-Key: $apiKey" ` -o ".\$jobId.zip" ` "https://voiclone.it/api/v1/jobs/$jobId/download"

Example: developer API request

These examples target the production domain. Replace API_KEY with the secret displayed once when you create an API key.

Optional post-processing

Apply post-processing

Post-processing profiles Radio

curl -X POST "https://voiclone.it/api/v1/generate" \ -H "X-API-Key: API_KEY" \ -F "sample=@/path/to/example.wav" \ -F "text=The airdrop has been dispatched!" \ -F "duration=20" \ -F "output_format=wav" \ -F "timesteps=10" \ -F "count=1" \ -F "apply_post=true" \ -F "client_request_id=generation-$(date +%s)"

Responsible use and AI-generated audio

Do not use the service for fraud, harmful or deceptive impersonation, harassment, unlawful disclosure, or deceptive content. Lawful parody, satire and fiction are permitted when they are not misleading or harmful and are disclosed as required.

Generated audio is synthetic. Select the intended use, follow applicable disclosure rules, and do not present generated speech as a real person’s statement without the required consent and disclosure.

The interface’s disclosure checkbox is a product safeguard, not a guarantee that a particular use is legally compliant.

Simple Voice Clone - TTS

Voice Prompt

Reference Clip

Log in

Buy credits

How packs work

Choose by workflow

Account history

Use credits when you generate

Generated files

More credit packs

Most popular

Dashboard

Completion time

Estimate vs final cost

Job status

Docs

Available endpoints

API keys for integrations

Generation request for API v1

Radio

Check a job and download its ZIP

Example: developer API request

Responsible use and AI-generated audio

Jobs

Account

Credit summary

API key

Change password

Contact