Video, image, and audio results

Where the result shows up

Every tool writes its result back to the same task. Open the task on the home page when the status is completed. Video and image jobs offer a preview and a download. Audio to text and image to text put the extracted words on the task so you can copy them. Voice change returns an audio file you can play from the task.

The file is stored with that task for a short time. The archiving guide explains why you should download a copy instead of treating the task list as permanent storage.

Video settings

Text to video asks for a length and a quality. The qualities are low, 576p, sd, medium, 720p, high, and 1080p. The form starts on low, and the optional generated audio track starts unchecked. A reference photo is optional. The prompt can be up to 1000 characters.

Photos to video does not invent footage from a prompt. It builds one slideshow from the photos and clips you upload. An audio track on that form is optional. Lipsync and face change also return a video: lipsync from a face photo plus speech, face change from a source face placed onto a target photo or a short target video. Target videos for face change need to stay under 30 MB and 60 seconds.

Images, speech, and transcripts

Text to image returns a picture from your prompt. A reference photo is optional, and you pick an output size on the form. The prompt limit is 1000 characters, the same as text to video.

Voice change returns speech audio. You either type the words or upload a speech file. One of those is required. A reference clip, if you add one, must be longer than 9 seconds. Male and Female presets and library voices are not measured that way.

Audio to text and image to text return words, not a new picture or clip. Clear speech and a sharp photo of printed text produce a more usable transcript than a noisy recording or a blurred screenshot.