Avatar workflow / Evidence guide
How to create a two-person talking avatar: voices, positions and timing
A two-person Avatar uses one image showing two people and a separate voice track for each side. It is not the same as a single portrait reading a script containing two names. Spatial assignment and speaking order are the important controls.
Key takeaway
Show both faces clearly. Assign audio by left/right position, choose sequential or simultaneous speech, check the combined duration and quote, and review which mouth moves during each voice.01 / Avatar workflow
1. Put both speakers in the same image
Use a landscape image with two clearly separated faces and visible mouths. Avoid overlapping faces or having one speaker much smaller than the other. Make sure you hold permission for both likenesses, not just for the photograph as a whole.
Prepare a clean audio track for the person on the left and another for the person on the right. Distinct synthetic voices can make a fictional conversation easier to follow. The editor does not infer the correct speaker simply from names written in a script.
02 / Avatar workflow
2. Assign voices and choose the speaking order
In Avatar, select 2 characters and assign each recording to the matching image position. Choose Left → right, Right → left or Together. Sequential speech uses the combined audio duration; simultaneous speech uses the longer track. The current maximum is ten minutes of the resulting audio timeline.
Start with a short exchange so you can check assignment before producing a longer dialogue. Review the displayed quote after selecting Fast or Quality. Voice generation is separate from the Avatar charge; uploading existing authorized audio does not require generating new speech.
03 / Avatar workflow
3. Watch the original two-speaker example
The example below uses fictional people and original synthetic narration to ask what makes a city feel like home. It was made in Quality and is a first-party, credit-funded demonstration. Inspect the detailed record for the exact lines, timing, settings and original media.
Check each turn separately: the intended face should move while its assigned voice speaks, with no unexpected talking from the listener. Overlapping speech is harder to inspect and can make a public explanation less clear.
04 / Avatar workflow
When to use dialogue Avatar instead of Full Video
Choose Avatar when the central requirement is visible faces speaking to supplied audio. Choose Full Video Creator when the goal is a sequence of illustrated scenes, narration, captions and optional scene animation. These are different production workflows; a long storyboard export is not a continuous talking-head performance.
Download and inspect the whole MP4 before using it in a lesson, client explainer or marketing piece. Do not portray fictional participants as real interviewees or invent endorsements.
05 / Avatar workflow
Free access, costs and commercial use
Start without a subscription using free welcome or earned credits; paid credits are optional. If credits are insufficient and a fresh check confirms no surveys are available, eligible accounts may receive automatic lower-priority processing. It requires available free funds and must fit the existing daily subsidy limit. It does not promise unlimited free avatars of any duration.
Fast: 45 credits per billed second. Quality: 135 credits per billed second. Round audio duration up to whole seconds with a 3-second minimum. A 10-second avatar costs 450 credits in Fast or 1,350 in Quality; 30 seconds cost 1,350 or 4,050 credits respectively. Both include the 1080p-target upscale. Quality remains the default; Fast is an explicit lower-cost choice, not a free-processing switch. Creating the input image or voice separately has its own quote. The editor confirms the current price before submission.
Donatello permits commercial and monetized use of Avatar outputs under the Terms, including client explainers, lessons and marketing. You must have permission for the image, audio, voice and likeness, and comply with applicable provider terms. This is not a guarantee of copyright, exclusivity or freedom from third-party claims. Do not impersonate someone or present a synthetic presenter as a real testimonial.
Verification
Sources and current facts
Product conditions can change. Donatello's dated Product Facts and binding Terms take precedence over this explanatory article if a material detail is updated.