What is bits per sample in audio?

[Anonymous543]

And in YouTube videos they said 8 bit were used for classic games like Mario contra etc ,so in game their audio was sounded good ,and in modern games like subway surfers ,free fire etc ,they use 24bit audio depth , why they needed 24bit if sound is good at 8 bit too??

When i did experiment with my ac3 audio track... converted 32bit per sample to 8 bit per sample ,it added some noise in it why that happened?
Because in Nintendo games there was no noise in game audios

[davexnet]

Because of quantization errors
https://www.mixinglessons.com/bit-depth/
https://www.izotope.com/en/learn/digital-audio-basics-sample-rate-and-bit-depth.html

[hello_hello]

To be precise not apps but apps developers, usually those without in-depth knowledge about DSP - 32bit float is perceived by many as convenient, universal format that provide more than enough dynamics and at the same time (largely falsely) precision (barely 25 bits) where dynamics and precision are not the same thing - float 32 seem to be convenient as in common misconception it will remove from developer responsibility for signal path (this is based on false assumption that 32 bit float means high quality audio where 32 bit float is frequently insufficient and may lead to nasty problems...)

So what's the alternative to 32 bit float?
25 bits worth of precision doesn't seem too stingy.

[hello_hello]

I watched few youtube videos on bit depth...in that they are saying more bitdepth means more tunes can be converted to analog to digital

8 bit = 2⁸ =256 tunes can be converted
16 bit = 2¹⁶ = 65536 tunes can be converted
24 bit = 2²⁴ = 16777216 tunes can be converted

I don't know what they mean by "tunes" but other than that it's exactly what I said in post #2.

I don't know it's correct info or not but there's something quantization too idk anything about quantization...

<sarcasm> Maybe I should have explained what "quantization error" means. </sarcasm>
Admittedly I edited the post to add it shorty after submitting it, so maybe you missed it, but quantization error is explained in post #2 as well.

[Anonymous543]

I watched few youtube videos on bit depth...in that they are saying more bitdepth means more tunes can be converted to analog to digital

8 bit = 2⁸ =256 tunes can be converted
16 bit = 2¹⁶ = 65536 tunes can be converted
24 bit = 2²⁴ = 16777216 tunes can be converted

I don't know what they mean by "tunes" but other than that it's exactly what I said in post #2.

I don't know it's correct info or not but there's something quantization too idk anything about quantization...

<sarcasm> Maybe I should have explained what "quantization error" means. </sarcasm>
Admittedly I edited the post to add it shorty after submitting it, so maybe you missed it, but quantization error is explained in post #2 as well.

[Attachment 62774 - Click to enlarge]


Quantization error may be that flat horizontal line between two samples where audio gets noisy while playback??

[jagabo]

This is a whole area of computer science:

https://en.wikipedia.org/wiki/Data_compression

Lossless methods compress by removing redundancy. The output upon decompression is exactly the same as the input that was compressed. But most real world data doesn't compress by much. Something like audio usually gets compressed to somewhere between 1/3 and 2/3 the original data size.

Lossy methods only approximate the original data. For example, the Discreet Cosine Transform is often used:

https://en.wikipedia.org/wiki/Discrete_cosine_transform

The output is not exactly the same as the input. But the amount of compression can be much higher. Generally, the more you compress, the larger the errors, the worse the quality.

[pandy]

So what's the alternative to 32 bit float?
25 bits worth of precision doesn't seem too stingy.

This depends what are you doing with signal - sometimes you need more bits - for example IIR filters may require 40..56 bits at least to not oscillate - so based on this you may say that double float is probably sufficient for most common audio processing but if you are clever then perhaps 32 bit integer will be sufficient (imagine product with 32 bit integer and product with 64 bit float - which one will be for example more energy efficient - important for portable).

25 bit precision is more than enough for playout but there is no DAC's accepting natively 32 bit float so at some point you need to convert 32 bit float to integer anyway...

[pandy]

Quantization error may be that flat horizontal line between two samples where audio gets noisy while playback??

Quantization error in simplistic definition is a difference between real analog signal value and signal value after quantizer in digital domain - for example imagine video signal (audio use simply bigger numbers so i will use video) - 8 bit quantization, black level is 16 and white level 235 - this is very common - if there is black then it will be no quantization error and value 16 will be equal to analog black, same for white but frequently we need gray level 50(%) i.e. exactly between black and white - and in 8 bit you can't have 50% gray - as 50% gray need 125.5 value in digital domain but 8 bit digital domain value can be or only 125 or only 126 and this is your quantization error...