While neither product is on the market, it’s still fun to compare two products based on their public demos. Both systems are designed to product TTS functionality / voice insertion after having analyzed previous samples. I was amazed by VoCo when it was announced in the fall. See Is #VoCo Loco?
The original Adobe Max keynote with VoCo is here:
A colleague pointed me to a great article that was put on on VoCo a few days ago. You can check out the video here:
For Lyrebird, you can check out their demo here:
On the face, VoCo just kills Lyrebird. A few ways:
- The sampling rate for Lyrebird seems to be really low
- The intonation still sounds robotic and forced
- We don’t know anything about the tools behind the software
It could be that we’re not comparing apples to apples. VoCo requires 10 minutes of samples and the demos show the insertion or changing of one word where Lyrebird might be entirely a TTS engine from a single sample.
Can’t wait to try a bake off and a Turing Test of these two.
A small public service announcement — DO NOT sign up at the site vocabeta dot com. It seems to be a fake site for capturing emails.