Overview
Combine one portrait with one speech recording to make a talking head video, then download the MP4 with audio. Sign in and stay connected while both files upload for online generation. Choose an available quality option that fits your recording, then review the full video: animation can change facial details or look out of time with the voice.
AI Talking Head combines one portrait with one speech recording to generate a video with audio. The video length follows the supplied speech; excess generated padding may be trimmed. The result is an MP4, but technical checks don’t guarantee accurate lip timing or unchanged facial details.
When to use
- Use AI Talking Head when you have a portrait and a speech recording and want a talking video without filming a new video of the person. Both files are required; this is not a text-to-speech or batch-animation workflow.
- It can suit a short introduction, lesson, or announcement. Start with a clear, well-lit, front-facing portrait with the mouth visible, plus a single-speaker recording with little background noise. These recommended inputs don’t guarantee natural movement or accurate lip timing, so review the result before sharing.
How to use
Open AI Talking Head and sign in. Under Face Image, choose a clear, front-facing portrait with the mouth visible. Accepted formats are JPG, JPEG, PNG, and WebP; the displayed limit is 20 MB. Use a portrait you have permission to animate.
Under Audio Track, add a single-speaker recording in MP3, WAV, OGG, M4A, or AAC. The displayed limit is 100 MB. Use understandable speech with little background noise, and check its length against the quality options. If you selected the wrong file, use Remove beside that input and choose another.
Under Video quality, choose an available option that fits the recording: Standard supports up to 15 seconds of audio, and HD supports up to 30 seconds. If the recording is too long, trim it or choose an available tier that fits. Leave Expression Style at natural; although natural, expressive, and subtle are shown, no animation change is documented for this selector on the current online path.
With both files loaded, select Generate AI Talking Head. Keep your internet connection during upload. Generation is online and can continue in the background; completion is reported through the bell. The workspace shows an estimated processing time of 5–10 minutes, but startup or busy periods can take longer.
Play the whole result with sound. Check mouth timing, facial appearance, head movement, the full speech, its beginning and end, and the downloaded dimensions. Animation can change facial details, so treat the result as an AI-generated animation, not a record of the pictured person actually speaking. Select Download to save the MP4, then inspect the saved file as well as the preview. Keep the original portrait and recording.
Settings reference
| Setting | Options | Default |
|---|---|---|
| Video quality | standard: Standard detail, up to 15 seconds of audio; dimensions adapt to the portrait. hd: Higher detail, up to 30 seconds of audio; dimensions adapt to the portrait. | standard |
| Expression Style | natural: Visible natural choice; no documented animation effect in the current online path. expressive: Visible expressive choice; no documented animation effect in the current online path. subtle: Visible subtle choice; no documented animation effect in the current online path. | natural |
Formats & limits
| Detail | Requirement |
|---|---|
| Input formats | JPG, JPEG, PNG, WebP, MP3, WAV, OGG, M4A, AAC |
| Output formats | MP4 |
| Input codecs | Decodable JPEG, PNG or WebP portrait plus readable MP3, WAV, OGG, M4A or AAC audio containing an audio stream. Extension alone does not guarantee decodability. |
| Output codecs | MP4 video containing audio; no codec selector is exposed in this workspace. |
| Bit depth | No user-selectable bit depth or HDR-preservation guarantee is provided for the generated MP4. |
| Dimensions | Output dimensions adapt to the portrait and selected quality. No fixed width/height or frame rate is promised. |
| Technical requirements | Supply one portrait and one audio file, each readable and at least 1,024 bytes., The audio duration must fit the selected available tier., The photo picker accepts 20 MiB and the audio picker accepts 100 MiB; these are separate limits. |
Access & cost
| Detail | Requirement |
|---|---|
| Sign-in required | Yes |
| Internet connection required | Yes |
| Limit scope | Talking Head dual upload: portrait up to 20 MiB and audio up to 100 MiB, displayed as 20 MB and 100 MB. Standard audio is limited to 15 seconds; HD to 30 seconds. |
| Plan limits | Use the current quality options and requirements shown in the workspace. The two upload limits and tier audio durations apply to this workflow; no additional plan-based duration is established. |
Examples & use cases
A presenter preparing a brief spoken introduction.
A short presenter introduction
They have a clear, front-facing JPEG portrait and a 10-second MP3 recording. They want a talking video for an introduction and need to check whether the mouth movement matches the speech.
JPEG to MP4
- Video quality:
- standard
- Expression Style:
- natural
- Use the JPEG under Face Image and the MP3 under Audio Track; both formats are accepted. The front-facing portrait keeps the mouth visible for review.
- Choose Video quality Standard because 10 seconds fits its 15-second audio limit. Leave Expression Style at natural because changing that selector has no documented animation effect on the current online path.
- Watch the full introduction with sound. Check mouth timing, facial appearance, and the beginning and end of the speech.
Expected outcome
An MP4 with audio to review and, if suitable, download. Check the saved file and its dimensions before using it.
An instructor preparing a short lesson segment from a recorded voice track.
A longer lesson segment
They have a clear, well-lit PNG portrait with the mouth visible and an 18-second WAV recording. Standard’s 15-second limit is too short for this recording.
PNG to MP4
- Video quality:
- hd
- Expression Style:
- natural
- Use the PNG portrait and WAV recording; both formats are accepted. The visible mouth and understandable speech make the result easier to review.
- Choose Video quality HD because its audio limit is 30 seconds, which fits the 18-second recording. Keep Expression Style at natural rather than expecting the other visible choices to change the motion.
- Play the whole lesson segment with sound. Check mouth timing, the spoken words at both ends, facial appearance, and the downloaded dimensions.
Expected outcome
An MP4 with audio for review. Inspect the saved video before adding it to the lesson.
A team member preparing a spoken announcement from an approved portrait.
A short announcement from a WebP portrait
They have permission to use a front-facing WebP portrait and a 12-second AAC recording with one speaker. They want a talking video and need to check for timing mismatches before sharing it with colleagues.
WebP to MP4
- Video quality:
- standard
- Expression Style:
- natural
- Use the WebP portrait and AAC recording; both formats are accepted. Confirm the mouth is visible and the speech is understandable.
- Choose Video quality Standard because 12 seconds fits its 15-second limit. Leave Expression Style at natural because the current online path has no documented animation effect for that selector.
- Watch the whole announcement with sound, paying particular attention to mouth timing and head movement. Review the downloaded MP4 as well as the preview.
Expected outcome
An MP4 with audio to inspect before sharing. Check the saved file, its dimensions, and whether facial details have changed.
More ways to use it
- An instructor making a short course introduction.: They have a portrait and a 20-second speech recording and want a talking-head-style introduction without filming a new video. Choose HD, which supports up to 30 seconds of audio, and generate an MP4 with audio. Review the full clip for mouth timing, facial appearance, movement, and the beginning and end of the speech. Video quality: hd, Expression Style: natural.
- A team member preparing a spoken update.: They have a 12-second single-speaker recording and a front-facing portrait they have permission to animate. They want a short talking video to review before sharing with colleagues. Choose Standard, which supports up to 15 seconds of audio, and generate an MP4 with audio. Watch the full saved file and check speech timing, facial appearance, and movement before sharing. Video quality: standard, Expression Style: natural.
Troubleshooting
The recording is rejected as too long for the selected quality option.
Try this: Trim the recording to 15 seconds or less, or choose the available HD option if the recording is no longer than 30 seconds.
What to expect: Standard supports audio up to 15 seconds and HD supports audio up to 30 seconds. Audio longer than the selected tier’s limit is rejected.
The voice is clear, but the mouth movement looks out of time.
Try this: Use a clear, front-facing portrait with the mouth visible; avoid an obstruction or an expression that makes the mouth hard to assess.
What to expect: A clear portrait is a recommended input for reviewing mouth movement, but it doesn’t guarantee accurate synchronization. Watch the entire result, especially during fast speech or unusual sounds, before sharing.
Privacy
Online processing uploads files. Completed results may remain available for later download. See Privacy Policy for current file-handling and retention details. Check the processing mode before starting.
Frequently asked questions
Why are no quality options available?
The tool loads the currently available options when you open it. If they fail to load, refresh and try again. Don’t start a job until you can check its audio limit. Standard supports up to 15 seconds of audio and HD up to 30 seconds; choose only an option currently offered in the workspace.
Why can lip movement look wrong even when the voice is clear?
A portrait’s angle, an obstruction over the face, or its expression can make mouth animation harder to assess. Fast speech and unusual sounds can also expose timing errors. Try a clear, front-facing portrait with the mouth visible, then review the whole clip; these choices don’t guarantee accurate lip timing.
How do I make a talking head video from a photo?
Add one readable portrait under Face Image and one readable speech recording under Audio Track. Choose an available Video quality option that fits the recording, then select Generate AI Talking Head. Play the complete MP4 with sound and check its timing and appearance before downloading or sharing it. See How to use for the full workflow.
Can I create a talking-head video from a photo and my own audio?
Yes. The workflow combines one portrait with one speech recording. Audio Track accepts MP3, WAV, OGG, M4A, and AAC. Use a portrait and recording you have permission to use, then review the result for timing and changes to facial details.
What image and audio formats can I upload?
Face Image accepts JPG, JPEG, PNG, and WebP. Audio Track accepts MP3, WAV, OGG, M4A, and AAC; FLAC isn’t offered by the audio picker. A supported extension alone doesn’t ensure a file can be read, so both inputs must be readable.
How long can the speech recording be?
Standard supports up to 15 seconds of audio, and HD supports up to 30 seconds. If your recording exceeds the selected option’s limit, trim it or choose an available option that fits.
What video format and dimensions does it export?
The downloadable result is an MP4 with audio. Its dimensions adapt to the portrait and selected quality; custom width, height, and frame rate aren’t available here. Check the downloaded file’s dimensions before using it.
Can I customize the portrait animation?
The visible Expression Style choices are natural, expressive, and subtle, but the current online generation path doesn’t forward that selection to the animation request. Keep natural selected rather than expecting a documented change in motion.
Can I use my own face and voice recording?
You can provide a portrait and a recording, but use them only when you have the necessary rights and permission, including permission to animate the pictured person. The workspace labels the result AI-generated.
Does the result always preserve the person’s face and match the voice?
No such result is guaranteed. AI animation can change facial details or produce movement that doesn’t match the recording. Watch the full clip with sound and check mouth timing, facial appearance, and movement before sharing.
What kind of portrait should I use?
Choose a clear, well-lit portrait of one person facing forward, with the mouth visible. This is a recommended input, not a guarantee of natural motion or accurate lip timing.
Does this tool turn a written script directly into a talking video?
No. AI Talking Head requires both a portrait and an audio recording; it isn’t a text-to-speech workflow. If you need speech made from text, Turn your script into spoken audio generates a WAV you can listen to and download; WAV is accepted under Audio Track.
How long does generation take?
The workspace shows an estimated processing time of 5–10 minutes. Startup or busy periods can take longer, so treat that as an estimate rather than a deadline. Processing can continue in the background, with completion reported through the bell.
Can I use any language or accent?
This workflow uses a supplied audio recording rather than a script with a language or accent selector. It accepts MP3, WAV, OGG, M4A, and AAC. A language-support list isn’t specified, so check the speech and mouth timing in the generated clip before sharing.
Why is my file rejected even though its format is supported?
A supported extension doesn’t guarantee that a file is readable. The portrait must be decodable, and the audio must contain a readable audio stream with measurable positive duration. This workflow also rejects missing inputs and files smaller than 1,024 bytes.
What are the upload size limits?
The displayed limit is 20 MB for Face Image and 100 MB for Audio Track. These are separate limits for the portrait and recording.
Do I need to sign in, and does the tool work offline?
Sign-in and an internet connection are required for this workflow. The portrait and audio are uploaded for online generation.
How are my uploaded portrait and audio handled?
Online processing uploads files. Completed results may remain available for later download. See Privacy Policy for current file-handling and retention details. Check the processing mode before starting.
Can I make many talking videos at once?
This workflow takes one portrait and one speech recording to generate a talking video; batch animation isn’t supported by the documented workflow.
Can I use a generated talking-head video for commercial work?
The documented workflow doesn’t establish commercial-use rights. Use portrait and speech material only when you have the rights and permission to use them, including permission to animate the pictured person, and check the applicable rights for your project.