So the problem is that white noise doesn’t compress very easily.
Compression algorithms are generally designed to reduce noise; if you have something that’s extremely noisy it’s really hard to compress because that’s not what the algorithms were designed to do.
This means that these podcasts take up more space, which means they use more bandwidth than an equivalent non-white-noise solution.
A middle ground would be banning these “podcasts” and then having a white noise generator built into the app. The white noise generator would run locally on your device (very easy to make white noise) and wouldn’t cost any bandwidth at all.
They should just do it procedurally on the fly.
because after all there was probably some algorithm that generated those all these and saved to a file. Just cut the file…
I’ve wondered for a while now why so few devices seem to generate it on the fly. Even Google Home and Alexa devices seem to play a 1 hour long file that fades out and in. The older, standalone “sound spa” units played a loop a few seconds long, which bothered some listeners who could hear a pattern due to the loop (maybe due to compression artifacts). I imagine it’s probably just computationally more expensive to generate it on the fly, rather than playing a file, but I also suspect that it’s also just companies pushing out the minimum viable product, and looping an audio file is easiest - especially if the device is already designed to play music, or other audio files like “ocean waves” and “babbling brook.”
To this end, one of the pixel phone launches had a website with a really good background idea, which was looping so nicely on the website that I thought I’ll download it and use it myself but it loops so badly on their own bedtime sounds in the clock app and ytm that it’s just a terrible experience, I never tried it in VLC though
I can totally hear a pattern overlaying the white noise sound in my kid’s white noise machine. The sound of the pattern varies based on the level of battery charge, as far as I can tell. I thought it was some kind of unintended noise coming from the circuitry (like maybe bad caps). Given that the underlying white noise sound doesn’t seem to vary based on the state of charge, I am still not sure that the sample length is what causes the pattern, but now you’ve got me super curious to tear the damn thing apart to test the caps.
It’s not just actual “white noise.” It’s many kinds of background noise like nature sounds, etc. Has to be recorded and often edited.
It’s a legit product that makes sp0tify more valuable. They should embrace it but they’re fucking morons who hate both their artists and their audience.
Another solution then: automatically download/cache a user’s most frequently played tracks. I know downloading is a premium feature or whatever but they should consider it if it would save them money.
Not just frequently played tracks. Sell me the MP3 or WAV or FLAC file, give most of the money to the artist, and let me transfer the files to any device I want. Anything less is too restrictive.
What do you recommend to me so I can easily download and discover music that’s not spotify. You hate them but it seems they’re the top of the game right now
pandora has a really good recommendation algorithm. but for downloading? I recommend Deezer. you can download music with deemix-gui to download to mp3, wav, flac
Youtube music is pretty good on recommendations and its free with youtube premium so i think its a vastly more valuable service. (Or just use music revanced and get it for free)
Allmusic is 100% my favorite place to find new recommendations. I get to actively hunt instead of being fed by an algorithm (though they have too many ads now).
If you absolutely must be a $p0tify zombie who is mentally unable to find recommendations literally everywhere else in our music-obsessed world then you can still buy the albums you like to actually support the artists instead of basically pirating through a big tech app.
For streaming services like Spotifiy, whitenoise does not use more bandwidth than any other podcasts, songs etc.
This is because they are usually using lossy compression which will compress the audio with a specified average bitrate, so e.g. 160 kbits. This means the final audio stream is about the same size no matter what content you’re listing to. This would only matter for streaming sites like Tidal that use lossless compression to deliver a identical bit stream. And even then, the best lossless compression algorithms only safe about half the bitrate, so worst case the audio stream would be twice the size.
Also with modern lossy audio compression algorithms like opus, people will struggle to hear any difference with most kind of audio (including white noise), even in A/B tests. With lossless compression you don’t have a identical bit stream to the source material, but with high enough bitrates it is good enough that people won’t hear a difference.
I agree though that a white noise generator in the app would be a nice feature to save some bandwidth.
So the problem is that white noise doesn’t compress very easily.
Compression algorithms are generally designed to reduce noise; if you have something that’s extremely noisy it’s really hard to compress because that’s not what the algorithms were designed to do.
This means that these podcasts take up more space, which means they use more bandwidth than an equivalent non-white-noise solution.
A middle ground would be banning these “podcasts” and then having a white noise generator built into the app. The white noise generator would run locally on your device (very easy to make white noise) and wouldn’t cost any bandwidth at all.
They should just do it procedurally on the fly. because after all there was probably some algorithm that generated those all these and saved to a file. Just cut the file…
I’ve wondered for a while now why so few devices seem to generate it on the fly. Even Google Home and Alexa devices seem to play a 1 hour long file that fades out and in. The older, standalone “sound spa” units played a loop a few seconds long, which bothered some listeners who could hear a pattern due to the loop (maybe due to compression artifacts). I imagine it’s probably just computationally more expensive to generate it on the fly, rather than playing a file, but I also suspect that it’s also just companies pushing out the minimum viable product, and looping an audio file is easiest - especially if the device is already designed to play music, or other audio files like “ocean waves” and “babbling brook.”
Because marketing doesnt want to ask engineering and think they’re brilliant when they come up with their own solution.
And when engineering tries to explain the downsides marketing gets mad like they’re being insulted.
To this end, one of the pixel phone launches had a website with a really good background idea, which was looping so nicely on the website that I thought I’ll download it and use it myself but it loops so badly on their own bedtime sounds in the clock app and ytm that it’s just a terrible experience, I never tried it in VLC though
I can totally hear a pattern overlaying the white noise sound in my kid’s white noise machine. The sound of the pattern varies based on the level of battery charge, as far as I can tell. I thought it was some kind of unintended noise coming from the circuitry (like maybe bad caps). Given that the underlying white noise sound doesn’t seem to vary based on the state of charge, I am still not sure that the sample length is what causes the pattern, but now you’ve got me super curious to tear the damn thing apart to test the caps.
It’s not just actual “white noise.” It’s many kinds of background noise like nature sounds, etc. Has to be recorded and often edited.
It’s a legit product that makes sp0tify more valuable. They should embrace it but they’re fucking morons who hate both their artists and their audience.
Fuck sp0tify a million times. I really hate them.
Another solution then: automatically download/cache a user’s most frequently played tracks. I know downloading is a premium feature or whatever but they should consider it if it would save them money.
Not just frequently played tracks. Sell me the MP3 or WAV or FLAC file, give most of the money to the artist, and let me transfer the files to any device I want. Anything less is too restrictive.
What do you recommend to me so I can easily download and discover music that’s not spotify. You hate them but it seems they’re the top of the game right now
pandora has a really good recommendation algorithm. but for downloading? I recommend Deezer. you can download music with deemix-gui to download to mp3, wav, flac
Youtube music is pretty good on recommendations and its free with youtube premium so i think its a vastly more valuable service. (Or just use music revanced and get it for free)
Bandcamp has recommendations
Allmusic.com has “similar artists” and browse-by-genre and -style (and mood and theme etc)
Everything you listen to on YouTube sends you down a new rabbithole
Discogs.com has recommendations
Allmusic is 100% my favorite place to find new recommendations. I get to actively hunt instead of being fed by an algorithm (though they have too many ads now).
If you absolutely must be a $p0tify zombie who is mentally unable to find recommendations literally everywhere else in our music-obsessed world then you can still buy the albums you like to actually support the artists instead of basically pirating through a big tech app.
For streaming services like Spotifiy, whitenoise does not use more bandwidth than any other podcasts, songs etc.
This is because they are usually using lossy compression which will compress the audio with a specified average bitrate, so e.g. 160 kbits. This means the final audio stream is about the same size no matter what content you’re listing to. This would only matter for streaming sites like Tidal that use lossless compression to deliver a identical bit stream. And even then, the best lossless compression algorithms only safe about half the bitrate, so worst case the audio stream would be twice the size.
Also with modern lossy audio compression algorithms like opus, people will struggle to hear any difference with most kind of audio (including white noise), even in A/B tests. With lossless compression you don’t have a identical bit stream to the source material, but with high enough bitrates it is good enough that people won’t hear a difference.
I agree though that a white noise generator in the app would be a nice feature to save some bandwidth.