Text to Audio and Text to Speech

What this looks like in mydubly

What you do

1. Paste a paragraph, or switch to Conversation and write each line on its own. A script can run for up to two hours of speech.
2. Set the language to speak. The language the text is written in is read from the words. The same language speaks the words as written.
3. Use one voice for a paragraph, or a different voice on each line. A paragraph can also copy a recording of 3 to 15 seconds. A longer sample keeps the first 12 seconds.
4. Confirm you have permission when you clone. Stock voices do not copy anyone.
5. Download the audio. To change a line or a voice, edit the script and speak it again.

What it costs

This uses dubbing credits, the same price as translating a video: 100 credits (10¢) for the two-minute minimum, then 50 credits for each extra minute of estimated speech. A one-sentence announcement still meets that two-minute minimum, so paste a whole script in one go rather than one line at a time.

Changing a line, a voice or the spoken language and speaking the script again is a new reading at that same price, including the two-minute minimum.

Voices

Studio voices cover clear narration, a bright conversational tone, a calm read, a low serious voice, news, storytelling and an energetic pace. Mood voices cover horror, bedtime, whisper, ASMR, sadness and anger, in male and female deliveries.

A paragraph speaks the whole script in the one voice you set. Conversation mode keeps each line in its own voice, so a dialogue stays a dialogue. Your recording is for a paragraph, when you have permission to copy that voice: one person, no music, at least three seconds. The sample is stored with the job and deleted with the rest of the reading.

These voices speak Chinese, English, French, German, Italian, Japanese, Korean, Portuguese, Russian and Spanish. The words can be written in any translation language. A recording can still be spoken in the wider set of dubbed languages.

Edit, re-process, download

The finished reading stays on this page. Hide the player, change any line or voice in the script, then speak it again. The new audio replaces the player.

Download the audio as M4A. There is no picture. Files stay on our servers for 120 minutes after the reading finishes, then they are deleted. The account keeps the opening words as the file name, the length, the languages and the credits.

What this is not

• It does not read a PDF, a webpage or a YouTube link. Paste the words.
• It does not keep background music, because there is no original recording.
• It does not lip-sync, and it does not make a video.
• A cloned voice is only as clean as the sample. Noise, music or two people talking in the sample will be in the copy.
• Very short text is billed at the two-minute dubbing minimum.