Like the infinite monkeys typing Shakespeare, but with audio instead.
If there was a program that created a series of sounds at random intervals, pitches, amplitudes, etc., how long would it take to produce an output that sounds like music, some sort of recognisable recording (e.g. a bell ring, a dog barking), or perhaps even a human voice?
Quite similar to the Library of Babble, which contains every single possible combination of letters. I don’t have an answer to your question exactly but you can try exploring the library or babble to get an idea
I feel like people here are arguing about technicalities, when the answer simply is that there is no way to know. If it is completely random, it could create a melody in the first 5 seconds. Or in 5 years from now.
You could only decide on a lower bound, i.e. for anything to be recognizable, you might need at least 100 ms of it, so it will take at least 100 ms to produce that.
But as soon as those 100 ms are over, all bets are off. At any moment, it could queue the appropriate sound that makes you recognize the previous seconds as some melody or sound.Build the program and let us know your findings
If it is truly random, where every outcome is equally likely, then Beethoven’s 5th could be the first thing produced. It could also take an infinite amount of time to produce.
Some would say that a generated random series of sounds is a form of avant-garde music. So it could be immediate.
It’s kinda impossible to answer, since there are infinite possible ways to randomly generate noise, for instance
- White noise is https://en.wikipedia.org/wiki/White_noise
- Pink noise https://en.wikipedia.org/wiki/Pink_noise
- Brownian noise https://en.wikipedia.org/wiki/Brownian_noise
(most random audio processes do not have a name). And if any audio is possible, like in these processes, you would never get an exact real song, but getting close (same as a real song but a little out of tune) would happen eventually.
I think a better analogy to infinite monkeys typing Shakespeare is randomly hitting the keys on a piano, though still, you would need to specify how the timing of the notes is randomized. If the lengths of the notes and the pauses are uniform and limited to standard lengths, eventually you would get, say Mozart’s Symphony no 5.
It’s also impossible to answer because pareidolia will cause different people to hear what they think is music or speech in random noise at different thresholds, dependent on the person.
not too long, I would think. animal collective seems to put out an album every few years.
I used to think this until I listened to Merriweather Post Pavilion while on a particularly large dose of psilocybin like a decade ago.
Absolutely incredible album. Never heard anything like it before or since.
haha, very much the same for me. i do appreciate some of their albums while not on drugs now, though.
Boards of Canada was another one I’d always listen too if I was on something. very different music style for AC though.
I’d heard of beer goggles, and now I’ve heard of psilocybin ears.
Not much sticks past maybe a week or two after a trip, in my experience, but I can listen to that album sober and appreciate it just as much. I just needed to be altered in order to truly give it a listen the first time
Can you still listen to it? I found i have a hard time with things like music and tv that ive seen high
oof!
Random intervals, pitches, amplitudes: essentially white noise? If I start generating it on my computer now I’d not count on it happening in my lifetime.
happening in my lifetime.
How long is that?
5
5 what? Decades? Years? Months?
4,3…
I have no idea how to even try to tackle this. Yet this is one of the best “no stupid questions” in a long while. Not only is it not stupid, it’s inspired. Complex yet impractical, thus perfectly suited to this sub. Bravo!
Depends on what’s random imo. Random amplitudes is just white noise, random pitches changing at random intervals could sound like music right away
About 10 minutes to get to Yoko Ono.
have you ever used that random book generator? no matter how much you click it, it’s never going to randomly generate something workable. there are way more unusable combinations of letters than usable ones. I think the same principle would apply. you’d just get string after string of random, discordant sounds.
no matter how much you click it, it’s never going to randomly generate something workable
Eventually you’ll click it enough to generate every written work ever produced. It will even produce a perfect narrative of your entire life, including the private moments nobody else could possibly know.
Not necessarily. There are an infinite number of outputs that aren’t those things.
Yes, in a span of now to infinity. It could take 13 weeks, or maybe 13 billion years. Maybe you’ve been clicking the button until entropy has leveled out, the final book of your life being the one written at the death of the universe. But your story hasn’t ended yet. You’re now a branch that needs to be pruned, existing in a space that doesn’t exist in any way we could ever conceive of…clicking the button. You got your hopes up 34 quadrillion years ago, but the final word was “their” instead of “there”. You’ll get their.
It’s been a while since I’ve studied statistics, but I think this would be classified as a Bernoulli distribution, where 1 represents a noise that sounds like music, voices, etc., and 0 represents a noise that does not sound like those. In these distributions, “How long…?” is a matter of chance and averages of trials, not an exact datum.
I think it is safe to say that OP’s question was lay speak for “what is the mean time to get to a result”. Other than that I don’t think you actually addressed the question.
Let me try to get it started:
Randomly generating music might be akin to password cracking. Cracking short or simple passwords can be very fast, while cracking long or complex passwords can be very long. The rate of password guessing also affects the time to get a result.
To calculate an answer, we need the following information:
- Guessing speed (how fast is each “song” generated and checked?)
- Minimum “song” length that needs to be generated
- Complexity of “song”: how many instruments (“voices”), resolution (are whole notes only ok, or do we need. Half or quarter notes?)
- Settle on some subjective definition of “song”. Is S.O.S. in morse code a “song”
You might be able to take a genre of music, and decompose the songs within to get some answers… I don’t have the time for that. Anyone want to take a stab at estimating the calculation?
OPs question is just any audio that strikes the listener as being a “real” sound. Doesn’t have to be long. Doesn’t have to be a song.
Because it just has to be “a” “real sound” i think there is an inherent measure of subjectivity. I might think a sound sounds like something you might not.
I think I’d approach this differently. I’d just pick a short time frame (maybe 0.5s) and generate 64kbs (PCM bitrate) worth of noise.
What percentage of those should have waveforms with any shape whatsoever within the domain of human perception. (What percent of random noise has the possibility of representation of a limited physical system interacting with the atmosphere in a way the human ear could perceive it)
Then, of that, subjectivity what percentage of those sounds “sound like a thing”.
Completely random sounds?
Technically, never. Physics is a bitch.
Interesting thought experiment!
How much compute you got?
Does it have to be a a full song? A snippet? We would also have to determine a “Turing test” style evaluation framework, and it had better be fast too - it cant be you and a pair of headphones determining success :)Anyway… So, with LOTS of compute, I suspect you could get a snippet or moment that resembles a song/voice, in relatively short time-frames, but this is only because true randomness is hard and we teach computers based on the past So patterns (snippets of voice or song), based on what the computers had previously learned from likely would emerge “quickly” - something like crypto mining, so applying large volumes of compute for extended periods of time
Now, in the spirit of what you are asking, the “infinite monkeys on infinite typewrites produce Shakespeare” or in this case, a song…
So, how long to produce a song/voice in a “truly random” scenario where an infinite number of thinking machines, never having heard a song or human voice … The answer is Undetermined. Infinity is weird
If we give it infinite time, then an infinite number of songs and plays will eventually be produced, but since we dont have infinite time, the most likely outcome (in my opinion) is that it would never produce anything we would all agree is a voice or song in any sort of human scale timeframes
Infinity is really really really big - did you know there are “bigger” and “smaller” infinities? This is the example that I found fascinating:
There are infinite whole numbers 1, 2, 3…Infinity within that set, there are infinite odd numbers, 1,3,5…inifnity also within that same set there are infinite even numbers, 2,4,6…infinity
Although all are infinite, the first infinity contains both the others, therefore is larger Mind blown when I first thought about that
Great question and fun to think about
