YouTube Shorts AI voice: a 59-second voiceover for about 5 cents

"YouTube Shorts AI voice" is on Google's autocomplete today (read 2026-10-07). Reported by HeyOrca, the Shorts editor now separates audio and video tracks, so a voiceover is a track you can swap without touching the picture. On Sume the voice comes from the text-to-speech endpoint, and the file is a normal audio artifact you can lay over any vertical video.

Size the script

Spoken English runs around 15 characters per second, which is a planning figure rather than a Sume measurement. At that rate a 59-second Shorts script is about 885 characters. At $0.0475 per 1,000 characters (read 2026-10-07) that is $0.042, rounded up to $0.05 for the job.

Make the file

Call POST /v1/tts-1.0/generate with transcript and a voice, then poll the job. Ask for WAV at 44,100 Hz if you plan to edit it, because a lossless file avoids a second lossy encode when the Shorts editor re-exports. Keep each request under the 20,000-character and 1,200-second limits.

curl -X POST https://api.sume.com/v1/tts-1.0/generate \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: shorts-vo-001" \
  -d '{"transcript": "Three habits that doubled my watch time.", "voice": {"id": "YOUR_VOICE_ID"}}'

Put it on the video

Two routes work. Upload the file straight into the Shorts editor as its own audio track, or render it onto the picture first with a Timeline 1.0 job at $0.10 per output minute, using the clip as the video layer and the voiceover as audio.url. Check the final length against the Shorts limit your account shows; trim the script, not the speech speed, if you run over.