So, shortly after making the last post, I found out that most of my themes broke. I didn't want to assume it was the bgm.bcstm files that I spent a short amount of time calculating, but of course it was. For some terrible reason, I wanted to imagine that the person telling me that 3.3MB was talking about MiB because that would be the easy way to read the sizes that Windows tells me. Instead, the people that said 3.3MB actually did mean megabytes and not the fancy mebibytes that literally everyone is used to.
Since I took the time to figure out most of the stuff already, and the ending number is actually pretty useful for predicting what sampling rate I'd probably have to use, I thought I would update the numbers.
As it is 3.3MB, that ends up being 3.21484375 MiB and our target .wav file size would be 11.20154616724738675958 MiB. Plugging in that fancy number ends up yielding:
sample rate (Hz) = ( 2867.59581881533 / time ) * 1000
Unfortunately, even though it IS a relatively decent ballpark, the bullshit in the last post was actually more useful and more accurate. This was probably because of introducing a number almost literally pulled out from a bag (the 0.287 ratio between bcstm to wav) so early into all of the calculations. So for this post, I thought I would go from the opposite end and use my already fairly bcstms to get even closer ballparks numbers.
The main ratio I'm looking for is still the ratio between the sample rate and time aka the 2867.595818...etc, but since I have slowly tried to find the optimal sample rates for at least a couple (to the nearest 50 Hz for most) I can start from there. So here is a quick table that shows the optimal numbers I got for the different themes I have.
As you can probably tell, the average ratio is about 2928.726 after removing the Haruhi outlier which is removed because although it does loop, it looping the entirety of the track rather than a subsection and thus possibly works differently. Since these are based off of actual concrete numbers without introducing a large amount of bullshit ( though honestly 0.287 isn't actually completely bullshit, it just wasn't close enough for me). The ratio is closer to my previous post's conclusion, which is why my ballpark numbers were so conveniently close to what I considered "optimal" in this post. Using the number way up top from this post pretty much guarantees that it will be within the limit by at least a couple hundred Hz, but using the average I just got would probably ballpark it within the nearest 100 Hz which is probably indistinguishable from one another auditorially.
Pretty much to summarize:
sample rate (full loop) = 2,922,920 / time (sec)
sample rate (sub loop) = 2,928,726 / time (sec)
with the optimal sample rate plus or minus 50 Hz.
-- Edit 2023/04/29 --
I revisited the topic and came up with some more precise numbers that can be used.
https://gbatemp.net/blogs/3ds-theme-bcstms.19325/
Since I took the time to figure out most of the stuff already, and the ending number is actually pretty useful for predicting what sampling rate I'd probably have to use, I thought I would update the numbers.
As it is 3.3MB, that ends up being 3.21484375 MiB and our target .wav file size would be 11.20154616724738675958 MiB. Plugging in that fancy number ends up yielding:
sample rate (Hz) = ( 2867.59581881533 / time ) * 1000
Unfortunately, even though it IS a relatively decent ballpark, the bullshit in the last post was actually more useful and more accurate. This was probably because of introducing a number almost literally pulled out from a bag (the 0.287 ratio between bcstm to wav) so early into all of the calculations. So for this post, I thought I would go from the opposite end and use my already fairly bcstms to get even closer ballparks numbers.
The main ratio I'm looking for is still the ratio between the sample rate and time aka the 2867.595818...etc, but since I have slowly tried to find the optimal sample rates for at least a couple (to the nearest 50 Hz for most) I can start from there. So here is a quick table that shows the optimal numbers I got for the different themes I have.
Code:
Optimal (Hz) | Time (Sec) | Theme | Ratio
32120 | 91 | Haruhi | 2922.92
28450 | 103 | Mikuru | 2930.35
28450 | 103 | Itsuki | 2930.35
23800 | 123 | Kyon | 2927.4
41230 | 71 | Ryoko | 2927.33
24200 | 121 | Tsuruya | 2928.2As you can probably tell, the average ratio is about 2928.726 after removing the Haruhi outlier which is removed because although it does loop, it looping the entirety of the track rather than a subsection and thus possibly works differently. Since these are based off of actual concrete numbers without introducing a large amount of bullshit ( though honestly 0.287 isn't actually completely bullshit, it just wasn't close enough for me). The ratio is closer to my previous post's conclusion, which is why my ballpark numbers were so conveniently close to what I considered "optimal" in this post. Using the number way up top from this post pretty much guarantees that it will be within the limit by at least a couple hundred Hz, but using the average I just got would probably ballpark it within the nearest 100 Hz which is probably indistinguishable from one another auditorially.
Pretty much to summarize:
sample rate (full loop) = 2,922,920 / time (sec)
sample rate (sub loop) = 2,928,726 / time (sec)
with the optimal sample rate plus or minus 50 Hz.
-- Edit 2023/04/29 --
I revisited the topic and came up with some more precise numbers that can be used.
https://gbatemp.net/blogs/3ds-theme-bcstms.19325/
