A Halcyone Maker field guide
Make a talking character video with your own audio
Start with one short line and one character. You can hear your audio with the moving mascot before deciding whether to export with Plus. The example below was recorded in the Maker with Honey Bear.
Made together in the Maker. Press play to see Honey Bear move and speak in one recorded clip.
Read the transcript
Hello, I am Honey Bear. Let us make something together. We can pause, take a breath, and explain the next idea. Thank you for watching.
1. Prepare a short voice clip
Record one idea in your usual audio tool, or start with Play voice example in the Maker. For a first try, aim for about 10 seconds: an introduction, one point and a closing line.
In Clip → Voice, choose Audio file. The mouth responds to its volume, rather than matching individual phonemes. Manual audio stays in your browser and is not stored with the character. No recording? With Plus, choose Script instead: type up to 200 characters and Make the voice (1 AI credit) — the mouth and speaking gestures follow it. AI edit → Make a talking video can also generate speech from a reviewed line and save its source audio with a speaking take.
- Use a file your browser can play, such as MP3 or WAV.
- Leave natural pauses and listen for distracting background noise.
- Have permission to use the recording in your project.
2. Open the character and preview the voice
Use Try a speaking character below. For a fresh project, it opens Honey Bear in Clip → Voice with a short speaking take. If you already have a saved or recovered character, your work takes priority. Voice input is available for dress-up mascots.
Choose Audio file, or press Play voice example. Once the audio length appears, press ▶ Preview. Free users can preview up to 20 seconds and replay it. Nothing is recorded or purchased by previewing.
- Watch the mouth during speech and pauses.
- If the gestures end too early, choose Fit speaking gestures to audio.
- That button replaces the current performance. Keep a custom sequence if you prefer your own timing.
3. Frame the whole performance
In Clip → Frame, choose Square 1:1, Vertical 9:16 or Landscape 16:9. Check the character at its widest gesture, not just the resting pose. Zoom out if the hands leave the frame.
Choose a background for your destination: transparent for an editor that supports WebM alpha, greenback for chroma key, or solid for a self-contained clip. Changing the shape in Clip keeps your background and take length.
Fitted gestures cover up to two minutes; longer audio continues while the final pose holds. For a longer lesson, separate short takes are easier to check and assemble.
4. Export the voice and character together
Choose MP4 video for social posts and YouTube under Clip → Export. For a video editor, choose Video editor under Where will you use it? to save a green-screen MP4 you can key out. GIF has no audio track. On Free, pressing Download with a voice opens the Plus guide; Plus or Pro unlocks voice-driven recording and microphone input. Basic silent exports remain available on Free.
After upgrading or reopening the page, choose the Manual audio file again. Your character and studio setup can be restored, but Manual audio itself is not saved. Preview again, then press Download MP4 (or WebM) and let the take finish.
Play the downloaded clip before producing a longer take. MP4 exports directly with H.264 video and AAC audio. MP4 has no transparency; transparent areas are saved on white. Choose a solid background in Frame to preview it, or WebM if you need transparency. If MP4 is unavailable in your browser, save as WebM.
5. Reuse the same character for the next video
Save the character with your account to keep its look, performance and studio settings. Free accounts hold up to three characters; see Pricing for larger save limits.
After a Manual voiced export, Make another video with this character clears the current audio so you can choose another file. On a later visit, reopen your saved character, go to Clip → Voice and select the new audio or type a new script. Refit the gestures only if you want to replace the previous performance.
In AI edit → Make a talking video, Save take keeps the reviewed line, production plan and source audio with your saved Mascot. Reopen a saved take later to review and make another ZIP. Finished video, GIF and PNG files remain separate downloads.
If the first preview is not right
No sound: check the device volume and wait for the audio length to load. Try another browser-playable file if an error appears. Preview starts only after you press it.
Little mouth movement: compare speech with silence in your recording. A very quiet voice or constant background noise gives the volume-driven mouth less contrast to work with.
The end feels too long: check both your audio and performance length. A voice recording lets the longer of the two finish, so shortening only the narration may leave a longer authored performance.
The background looks black after export: the player may not display WebM transparency. Try the transparent-video guide below before changing your character.