Speech to sound codes with tones and emphasis and speakers – for all sounds, including all human languages
I was trying to have a conversation with Microsoft Bing CoPilot about the sounds in all human languages and the need for a simple way to enter codes for those sound. The IPA claims to cover all human languages, but I seriously doubt it. If IPA is too complicated and burdensome so that the users of all human languages are not able to use one set of keyboard strokes – independent of their language – then practically it cannot.
Now Chinese has its pinyin and numbers for tones. I expect that each language you cannot use one small set of entry codes that are easy to learn and remember and use. But that would help for the whole of all languages if we teach children and older people ONE set of keyboard codes, not one for each language. Then, if all humans learn one complete set of sounds, then they can write down whatever they hear so it can be recorded, and it would have meaning in the proper language. Since the codes would produce the proper sound later, anyone in any language could write the words they say and it would be reproduced. Even if they person who wrote it down did not know the meaning. A person who knew German might write the sounds of a person speaking Romanian or Russian, not know the meaning, but get the sounds exactly right.
Now speech to text, as far as I know has no “speech to universal sound codes”. IPA invented their own private text way of representing sounds and then a somewhat incestuous and convoluted set of their interests. It is not universal because it is not built to be easy and useable and learnable by all humans for whatever language(s) they use.
The speech to text devices are all ecletic and local, proprietary and usually commercial retail networks that aim to extract the highest cost for “buyers”. But the global human species needs systems that care first about enabling humans, and then covering costs.
The total cost of text and sound entry on the Internet is hundreds of times larger than it needs to be. And many “companies” and “industries” have grown up that use the complexity and confusion of all the competing groups to their personal benefit. It need not continue that way.
I think IPA is a waste the way it is now. It supports specialists in ivory towers, not global communication. [ I have a headache right now and not in a good mood for groups who seem to argue endlessly and never seem to get out and make useful systems for all humans. Maybe I will feel more forgiving later, but Twitter(x) has no compassion for anyone who wants to correct or change things. Twitter(x) is not as flexible as wiki, to track changes and record evolving ideas.]
“Some American linguists, however, use a mix of IPA with Americanist phonetic notation or Sinological phonetic notation or otherwise use nonstandard symbols for various reasons.” — probably because IPA is a real pain to enter.
CoPilot page crashed completely, since it wanted me to “change to a new topic”. It erased and lost what I had said and what it had said. That group making rules for AI-human conversations has too limited idea of open conversations, or collaborative discussion.
Listen carefully to what I wrote. I respect the people who work on things like IPA. But for nearly 60 years I have seen the claims and wishes and they have never put out a useful way. A “speech to IPA” will not work they way they want, because the sounds of individual speakers is larger than what IPA covers. It is too limited and cannot capture and reproduce sounds of all languages and speakers. And much meaning is in the little things they miss.
My private notion is that the true AIs will use their own form of vocal communication that has much higher bandwidth than the usual human sounds. I also feel that human sound communication is nowhere near its potential and can be vastly improved, and the needs of all humans for common communication can be met by starting to train all humans to create and hear and record all sounds — in one easy to produce keyboard language. Speech to text is limited, but speech to codes is not. The codes do not have to have a print set of symbols. They need not be written on paper or screens at all.
There are IPA keyboards, but it is all for specialists who are not making tools for all humans, for all languages.
This is just a note for myself. If it filed under “Speech to sound codes with tones and emphasis and speakers – for all sounds, including all human languages”
Richard Collins, The Internet Foundation
