AI voice-over from a timed script
Paste SRT captions or timed lines, pick a voice, and get a voice-over where each line starts at its timestamp. Add your video to get it mixed.
Paste SRT or WebVTT captions, or one line per sentence starting with its time, such as 0:05 Welcome back.
Voice sample
The voice is synthetic (AI generated). Only use text you have the right to publish, and do not present the voice as a real person or use it to mislead.
Uploaded and processed audio is stored on SoundHalo. Audio access expires after 30 days; expiry does not confirm storage deletion. Delete the stored files here before leaving this page. Speech providers have their own retention policies. Privacy policy.
An example script and its result
This SRT gives each line a start and an end time. The end time sets how long the line may last.
1
00:00:00,500 --> 00:00:03,000
Welcome to the SoundHalo product tour.
2
00:00:03,200 --> 00:00:05,800
Upload a recording and we check it before anything else.
3
00:00:06,000 --> 00:00:06,900
Then download the result. This is the real line report for that script with the Heart voice. The third line has a 0.9 second slot but needed 1.08 seconds even at 1.25x, so it was flagged instead of cut:
0:00.5 Welcome to the SoundHalo product tour.
Fits
0:03.2 Upload a recording and we check it before anything else.
Sped up 1.12x
0:06.0 Then download the result.
Runs 0.2 s past its slot. Shorten this line. What is checked before you download
- The script is checked as you type: broken timing lines, times out of order and empty lines are shown with their line number, and lines that look too long for their slot are marked before you spend any allowance.
- After generating, each line shows whether it fit, how much it was sped up, or by how many seconds it still runs over.
- Listen to the voice-over and the mixed file on the page before you download them.
How lines are fitted
Each line starts exactly at its timestamp. Silence at the start and end of each generated line is trimmed first. If the speech is still longer than the slot, it is sped up without changing the pitch, up to 1.25x. Beyond that, speech starts to sound rushed, so the line is left at 1.25x and flagged instead. A flagged line keeps playing past its slot and can overlap the next line, which is why the report asks you to shorten it.
Mixing with your video or audio
- Lower the original: music and background sound drop by about 15 dB while the voice speaks (16 dB in our test with a music bed) and come back between lines.
- Replace the original: the file gets the voice-over as its only sound. Useful when the original audio is a guide track you do not want to keep.
- Video files keep their picture and are returned as MP4. Most videos are not re-encoded; a few formats, such as some WebM files, are converted to H.264.
Limits
- No account: a short preview of up to 300 characters and 20 lines, with 1,200 characters a day shared by everyone on your network and capped by a site-wide free budget.
- Free account: up to 3,000 characters and 200 lines per voice-over, 20,000 characters a day.
- Pro: up to 20,000 characters and 500 lines per voice-over, 300,000 characters a day.
- A voice-over can be up to 30 minutes long without Pro, and up to 4 hours with Pro. Characters are counted across all lines and shared with the text to speech tool; voices from Giggy count each line as whole blocks of 20 characters, because that is how each request is billed. Videos and audio to mix with are limited to 100 MB, and mixing counts toward the daily cloud processing allowance.
Synthetic voice and acceptable use
The voices are generated by text to speech models. They are not recordings of the people their names suggest. Use the audio for narration, tutorials, product demos, accessibility and drafts. Do not use it to imitate a real person, to make someone appear to say something they did not, or for spam and fraud. Some platforms require AI generated audio to be labelled.
Related tools
FAQ
What happens when a line is too long for its time slot? +
The line is sped up, up to 1.25 times normal speed, which still sounds natural. If it still does not fit, it plays in full and is flagged with how many seconds it runs over, so you can shorten the text and generate again. Words are never cut.
Can I add the voice-over to my video? +
Yes. Add the video (or an audio file) in the optional field. You get the voice-over track and a mixed file. Choose whether the original sound is lowered under the voice or replaced. Video keeps its picture and is returned as MP4.
Which script formats work? +
SRT and WebVTT subtitle files, or plain text with one line per sentence starting with its time, such as 0:05 Welcome back. Plain lines run until the next line starts.
Is the voice real? +
No. The audio is synthetic, made by a text to speech model. Do not use it to imitate a real person or to mislead listeners, and label it as AI generated where your platform requires that.
Where is my script processed? +
Each line is sent to a text to speech provider: Lemonfox for its languages and Giggy for the others. The generated audio and any video you add are stored on SoundHalo so you can download them. Access expires after 30 days, and you can delete the files from the page.
What are the limits? +
No account: a short preview of up to 300 characters and 20 lines, with 1,200 characters a day shared by everyone on your network and capped by a site-wide free budget. Free account: up to 3,000 characters and 200 lines per voice-over, 20,000 characters a day. Pro: up to 20,000 characters and 500 lines per voice-over, 300,000 characters a day. A voice-over can be up to 30 minutes long without Pro, and up to 4 hours with Pro. Characters are counted across all lines and shared with the text to speech tool; voices from Giggy count each line as whole blocks of 20 characters, because that is how each request is billed. Videos and audio to mix with are limited to 100 MB, and mixing counts toward the daily cloud processing allowance.