Voice as in human speech (a.k.a. telephony).
For analog, the answer is easy: 3.4 kHz (AM, SSB).
For digital, we are currently (more or less) getting to the ultimate limit by becoming able to transcribe speech to text and then re-create at the other end the timbre of the voice automatically and in real time (compressing the text is a long solved problem).
This would require even less bandwidth than what conventional telegraphy would need (less, because of the compression), somewhere below 1 kHz (the exact value would depend on the compression ratio).
_________
Of course, there is a way to do it with even less bandwidth, by getting artificial intelligence to construct a minimal representation of the meaning and transmitting that.
There is a catch, though: what do we do if nothing is actually said, all "words, words, words" and no meaning (as diplomats and politicians do sometimes)¿
The latter reminds me of an anecdote from the European Parliament, where simultaneous translation is provided for all languages of the member countries of the European Union.
The one having the floor delivers an in joke that would be understood only by a dozen fellow university students.
The interpreter says, "The speaker is telling a joke; please, laugh."
No comments:
Post a Comment