Duration: 0:00

96 views

1

I have been searching for a very good TTS for years, and now it's finally found. Kokoro is probably 90% created by the Chinese with unethical means. Here are some reasons why:

1, The ultra-national rhetoric is typical of a brainwashed Chinese mainlander:
Original text: 中國人民不信邪也不怕邪，不惹事也不怕事，任何外國不要指望我們會拿自己的核心利益做交易，不要指望我們會吞下損害我國主權、安全、發展利益的苦果！
Translated text: "The Chinese people do not believe in evil, are not afraid of evil, do not provoke trouble, and are not afraid of trouble. No foreign country should expect us to trade away our core interests or swallow the bitter fruit that harms our country's sovereignty, security, and development interests!"

2, There are few other TTSs that are also appearing to be created by the Chinese with English and Chinese as their own speech option. If only two options of English and Chinese are given, then it’s almost safe to say that it’s created by the Chinese.

3, How can I be sure that this is unethical? The voice data trained from is probably from Youtube, the “public-domain audio”.

4, Kokoro is a Japanese term, but this is very typical of the Chinese mentality. They are hiding their own identity using someone they truly hate. It’s comical and sad. On the other hand, anything made in Japan is of a higher quality, so using a Japanese term makes sense.

5, It has 5 Japanese voices, but 8 Chinese voices. You do the math. One of the Chinese voices has a strong SiChuan Chinese accent. This would less-likely be something created by a Japanese person. The SiChuan accent is associated with sexiness because SiChuan girls are known for their beauty.

If Kokoro is created by the Chinese, then it’s deceptive and disgusting.

PS. It's running on CPU only, but the AMD 4500u is still having plenty of power. This would be much faster on a better GPU.

PPS. The UI has been modified by me to make it more usable, so what you are seeing only exists on my computer.

PPPS. There are actually many ways that you can run Kokoro. Locally may not be the ideal. I can literally use my old windows 7 to access Kokoro via the cloud, faster with cloud power.

PPPPS. MS natural voices are good enough for most case and it's very fast in comparison to AI voice.

PPPPPS. Other less accurate TTS options could be more efficient. I know at least 2 that are more than acceptable while proportionally faster than Kokoro, but Kokoro is overall the best in terms of quality and processing power needed, especially in English. In mandarin, other TTS or even MS natural is more than enough.