Voices
Record your voice to use it in your video.
Voice cloning is an Enterprise Plan feature.
The Voices section displays all of the custom voices available on your account.
Click on the Preview button to hear how a custom voice sounds.
Create a new voice
Voice definition
To create a new voice in Synthesia:
- Click on
New voicein the Voices section. - Enter the following information:
- Voice name: The name of the voice to be displayed in the video editor language selection.
- Language: The language of your voice, only English is supported for now.
- Gender: With which avatar gender this voice should be displayed.
- Click on
Start Recording
Voice recording
On the recording page, you must record at least 12 paragraphs of text. For each paragraph you should:
- Click on
recordto start your voice recording. - Press
Stopwhen done reading. - If the recording is not successful, select
Record again.
Recording outcomes
During your voice recording, once a paragraph is validated, you have one of the following outcomes:
| Outcome | Reason | Solution |
|---|---|---|
| ✅ Passed | Minimal background noise, recording almost matches the script | None needed - you're ready to go! 🎉 |
| ❌ Failed - Due to Noise | This outcome will occur if background noise detected | Re-record your paragraphs |
| ❌ Failed - Due to speech recognition | This outcome will occur if significant mismatch detected between the recording and the paragraph text (if accuracy is at least 70% then mismatching words will be underlined) | Re-record your paragraphs |
| ❌ Failed - Due to speaker | A mismatching speaker was detected between the consent and the paragraph | Re-record your paragraphs with the correct speaker |
| ❌ Error | The sample validation service, is either unavailable or failed for some reason (recording can be re-submitted without re-recording) | Recording can be re-submitted without re-recording |
| ❌ Timeout | This will be when the UI does not receive a validation response within a minute, you can resolve this with the recording being re-submitted | Recording can be re-submitted without re-recording |
Perfect voice score
The Voice Score and the number of paragraphs in their various states (i.e not recorded, validated, processing, failed) can be found at the top of the page.
Make sure your voice score is as high as possible in order to maximise the quality of your custom voice.
Submission and consent
After recording the required number of paragraphs, click on Submit your voice at the top right of the screen to start the submission process.
In order to validate the submission, the voice owner needs to record their consent accordingly. The consent contain a unique passphrase that is randomly generated.
Validated consent cannot be re-recorded and it is important that the passphrase recording has to exactly match the passphrase for the consent to be accepted.
Once your consent has been processed, click Generate and your voice will be ready in a few minutes.
Using a custom voice
Once your voice is generated you can use it in your videos like any other voices by going into the Custom voices section:
After you selected it once. The voice can be used in all 29 supported languages!
Voice cloning supported languages
Arabic, Bulgarian, Croatian, Czech, Danish, Dutch, English, Filipino, Finnish, French, German, Greek, Hindi, Hungarian, Indonesian, Italian, Japanese, Korean, Malay, Mandarin (Simplified/Traditional), Norwegian, Polish, Portuguese, Romanian, Russian, Slovak, Spanish, Swedish, Tamil, Turkish, Ukrainian, Vietnamese.
Updated 2 months ago