h a l f b a k e r y
i v n i n seeks n e t o
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
or get an account
Anyone who's used a mobile for voice calls may have
noticed that the background noise often sounds like speech
or chirps. The characteristics of speech presumably have
some features in common with birdsong and others not,
and speech compression is presumably optimised for
there are other species which produce
sounds somewhat speech-like but not identical, for
instance birdsong, whale song and sounds made by other
primates. Other species make more regular sounds, for
instance cicadas and crickets.
So, my idea is this: find a method of sound compression
which is lossy but optimised for species. It falls into two
parts. One exploits characteristics common to all sound
produced by animals, from cicadas to cockerels. The other
is tweaked according to the species, the aim being to
produce an output which the species concerned can't
distinguish from the real thing, with the option of choosing
compression optimised just for that species or for two
different species. This would be interesting and useful for
a couple of reasons. Firstly, it would illustrate differences
between how humans and other species hear, and perhaps
in some inverted way could also enable us to hear
ourselves as other species hear us - useful for training
ourselves to train dogs and the like. Secondly, it would
enable us to store large amounts of audio data for specific
purposes, notably birdsong as an audio file on an alarm
clock, for transmission on digital radio or as an aid to
I haven't taken parrots into consideration here.
||Erm, no, the chirrups are probably down to lossy reception / poor data bandwidth of voice codec data, and nothing to do with how much the user looks or sounds like a Toucan.
The poor-old voice codec is presented with dodgy data, and tries to provide a sound by mimicing vocal-tract charactistics to match the (bad or inaccuate) data presented to it. What you hear on the otherside sometimes hear sounds bird-like.
One of the trickiest things for a good voice codec to do is to reliably and accurately encode phonemes from all languages from around the globe (some languages include clicks, buzzes and sounds unlike any European languages - so codecs developed and only tested in the West fail badly when used in other languages. Esperanto?) . The protocols must include enough redundancy to enable it to decode lossy data while keeping the data rate as low as possible.
Not merely baked, but positively roasted - The voice codecs are already optimized to species; The human species, as it has been discoved that humans have the money to invest in purchasing mobile phones. I know of no business model that specialised in marketing to any of our Aves class (mostly feathered) friends.
||You seem to have misunderstood me. The idea is
to use the voice codecs currently available but to
write new ones which work for the sounds made
other species, then optimise them either for the
species which makes the sounds concerned or for
the species which needs to hear them, or possibly
both. The chirps are just the seed of the idea.
||Concerning Esperanto, it'd probably be OK for a
codec which was good at optimising Czech, i'm
guessing - loads of palatals, fricatives and no
schwa. Not so much with Serbo-Croat because if i
remember correctly it's tonal (i probably don't).
||I have no idea about the glottal intricacies of Serbo-Croat, nor Pelican or Shrew. I am not a Doctor, and dont speak Alligator (though I have a smattering of Python).
So, the idea is not to allow birds to communicate reliably over a mass-access digital communications system. I stand corrected.
||No worries. Partly that, yes, also a way of hearing ourselves as others hear us and getting alarm clocks to play birdsong more effectively, and as a way of storing huge amounts of whale song on CD, compressing cricket chirps to tiny files and so on. In fact, maybe a whole rainforest of sounds could be sampled and put on a small water biscuit or something.
||// I haven't taken parrots into consideration here. //
||Shame on you. You should at least consider including the Norwegian Blue ("Beautiful plumage ....").
||I suspect that the first steps in this direction could be done
in hardware. Simply take a representative sample of
songbirds, compress them, and record the noise they make.
This short sound sample would probably contain, within it,
all the key phonetic elements of compressed birdsong.
||So their life flashes before their eyes, [MB]? The death rattle or dying scream of every creature recapitulates its life's vocalisations at high speed?
||"beady little eyes", as I prefer to think of them, but yes.
||Some bloke once recorded every note of the Ring cycle on top of the last one to produce a thunk which lasted a second or so.
||You could also work on a compression algorithm
specially for aliens from other planets. Itd save a lot
of space, the final frontier.
||Well, if he'd recorded each note *under* the previous one,
that would have been clever.
||The sounds during silence are added purposfully, and
part of the protocols (such as GSM). If you would
hear total silence, you would be sure your phone is
"broken". This was perceived at some time in 1993 or
4 when the first digital voice apps were being
developed. (We just had a reunion...)
||That sounds a little like a bug which became a feature, [pashute]. If i heard it on my 'phone, i'd probably wonder why i suddenly had a mobile when i haven't owned one for several years. However, that makes sense.