Mochi Factory · Blog

Can you hear the difference? A blind test of two AI narration models

Every scene in our HoneyStorm History videos is narrated by the same AI voice. ElevenLabs offers a cheaper, faster model that can use the same voice, so we asked: can the person who runs the channel tell them apart without knowing which is which? This is a small test (one clip, one judge).

How we tested

The test

The job: the same 17-second opening of the Dyatlov script, read by the same voice, with two ElevenLabs models.

Listen to A

Listen to B

Blind labelModelCost for this clipLength
AElevenLabs Multilingual v2$0.04917.2 s
BElevenLabs Flash v2.5$0.02516.1 s

The judge: "very close." When the judge can't tell, the cheaper one wins, so our narration moves to Flash v2.5, at half the cost with the same voice.

What we changed

History narration moves to ElevenLabs Flash v2.5 with the same voice: half the cost. A 20-scene video's narration drops from about $0.70 to about $0.35.

What we learned

  1. When it's "very close", take the cheaper one. Viewers won't notice a difference the channel's own maker can barely hear.
  2. Keep the voice, change the engine. Using the same voice meant the channel's narrator didn't change.
  3. Quality has to show up in the numbers. So we're running a follow-up: about 75 % of our next history videos are made with the cheaper models and 25 % with the original ones, picked at random. After enough are published, we compare watch time, retention and click-through by group. If the more expensive models don't earn its cost in viewer behaviour, it goes. We'll report the result here.

Want to see the result? The Dyatlov Pass is out now on HoneyStorm History.