I created an audio-visual using Google Cloud Speech API and Google Translate.
The visual concept is to "making art out of imperfection"
The challenge this time was to use AI to translate and visualize the lyrics into words.
Technically, the following is being done.
(1) Converts voice only extracts into Japanese text
(2) Translate the converted text into English
At stage (1), the accuracy is good, 80 or 90 percent.
In the process of (2), the accuracy is reduced.
Even so, for those who are deaf, it's a clue to the world of sound, and that's where the value lies.
People sometimes look for perfection in what they come into contact with.
But think about it.
There is no such thing as a clean road to begin with.
If we continue to make small steps, lost and struggling, the result will be a path that everyone can walk with peace of mind.
While AI can perform well in certain cases, it's unfortunately not universal in everything we do.
Today's AI works best when it accepts imperfections and combines various methods, and this law applies to many other things besides AI.
I hope you can feel the message of "making art out of imperfection" through this work.
Finally, I'm grateful to my wonderful friends who have helped me with this concept.
Lyrics&Poetryreading : hugvilla (hugvilla.jimdo.com)
Sound: Ikuko Morozumi (ikukomorozumi.com)
Visual: Ken-ichi KAWAMURA(twist-cube.com/vj/)
Character3D: TOMODACHI First machine(toky.jp)
Font: もじワク研究 (moji-waku.com)