AI voice cloning — generate speech in your own voice | NetRoom

Turn text into speech in your own voice

Record a short sample, enter a new line and generate audio with the voice from your recording. Use it to correct a line in a video or narrate the next part of a lesson.

How to add a voice sample and use it to narrate a new line in NetRoom.
1
speaker in the sample
Microphone
record in your browser
Audio file
use an existing recording

How it works

  1. Open Clone voice

    In Sound, select Speech and open the Clone voice tab. If the tab is missing, choose Clone your voice to switch to a compatible model.

    Sound / SpeechClone voice
    The speech editor is open on the Clone voice tab.
  2. Add a voice sample

    Select Record voice, allow microphone access and read a few sentences. You can also upload an existing recording or choose one from Files. Play the sample back.

    Synthetic demo voice Voice sample
    The first field lets you listen to the original recording.
  3. Check the words in the recording

    Transcribe exactly what you can hear in the sample. If the transcript field is optional, you can leave it blank. Each model’s requirements are listed below.

    Sample transcript
    The studio is quiet. I step up to the microphone and read a few lines. This is the sound of my voice.
    The transcript belongs to the recording. The new script has its own field.
  4. Generate the new line

    In the second field, try “We rewrote the ending. Good thing the mic is already packed.” Check the estimate and select Generate with my voice. Play the result, then download it.

    Fish Audio Drama 3 New line
    Play the finished line before downloading it.

Hear the sample and the new line

We created a synthetic voice for this example. Fish Audio Drama 3 then generated a different line from that recording. Listen for the timbre when the words change.

Voice sample Synthetic demo voice
The studio is quiet. I step up to the microphone and read a few lines. This is the sound of my voice.
New line Fish Audio Drama 3 Estimated cost to generate this line today: $0.00067.
We rewrote the ending. Good thing the mic is already packed.

What it does

Recording
Choose an excerpt with one speaker, without background music or overlapping dialogue. Use your own voice or a recording you have the speaker’s permission to use.
Sample formats
MP3, WAV, M4A, OGG, FLAC and WebM. Sample duration limits depend on the model.
Pricing
Character-priced models charge according to the length of your new script. For byte-priced models the language also affects the estimate. Check the estimate in the editor before each run.

What people use it for

Correct a voice-over

Changed the ending? Leave the microphone in its case. Use your saved sample to narrate just the sentence that changed.

Narrate a course

Use the same sample for successive chapters. Try a short line first to hear the timbre and pronunciation before generating a longer passage.

Try two introductions

Generate two openings from the same reference. Compare their rhythm and pauses, then download the version you want.

Voice cloning models and speech rates

Fish Audio S2.1 Pro

Voice sample
1–90 s
Transcript
Required
Speech generation
$0.018 per 1,000 characters

Billed by text size: a Latin character counts once, Cyrillic twice, Chinese, Japanese and Korean three times.

HeyGen Voice

Voice sample
1–180 s
Transcript
Optional
Speech generation
$0.018 per 1,000 characters

HeyGen Voice Turbo

Voice sample
1–180 s
Transcript
Optional
Speech generation
$0.0048 per 1,000 characters

Fish Audio Drama 3

Voice sample
1–90 s
Transcript
Optional
Speech generation
$0.012 per 1,000 characters

Billed by text size: a Latin character counts once, Cyrillic twice, Chinese, Japanese and Korean three times.

Frequently asked questions

Does my sample need a transcript?
It depends on the model. Use the transcript field for the words heard in the recording. Enter the new script for the output in the separate text field.
Can I reuse my sample?
Yes. The editor remembers your selection for your account in this browser. Remove clears the selection. To use the source on another device, save the original recording in Files.
Why was my recording rejected?
Check the model’s duration range and transcript requirements in the table. Play the file to make sure it contains audible speech from a single speaker. For a long recording, choose a shorter excerpt.

Record your first line

Start with one speaker and a single new sentence for your first take.