Realistic compression method
Regarding the bit rate, 192 kbps is sufficient for MP3.
A raw Wave file (44100Hz x 16bit x 2ch = 1411kbps CD sound quality) occupies approximately 42MB in 5 minutes, less than 5MB at 128kbps, more than 7MB at 192kbps, less than 10MB at 256kbps, but it depends on the user or if you really want to stick to the reproducibility of the treble range. If you want to welcome sound processing, you can also use lossless compression. With lossless compression, the more striking the sound (higher SPL and thicker highs), the harder it is to compress, but even with lossy compression, such sound tends to get distorted unless it has a high bit rate. The expression of sound quality is strongly influenced by the playback environment, and indeed if you play it at a small volume with a sloppy device, even 128 kbps is indistinguishable from lossy compression.
I also prepared a sample file (including the file explained in the next point). kanon_f_200.ape recorded the aforementioned fluent mix sound with a 200% Timidity ++ amp, and kanon_e_150.ape also recorded the aforementioned eaw-mix sound with a 150% Timidity ++ amp. Monkey’s Audio used version 3.99 and set the compression level to standard (high compression settings only slow down the process and don’t shrink much). The recording time of the fluid mix is approximately 3 seconds longer, probably reflecting the difference in the last sustain. The mixture of fluids is easier to gain sound pressure and can reach up to approximately 250%. Also, because the sound pressure is high, compression is not very effective. eaw-mix has noise-like ripples after 15 seconds (and 55 seconds), and 150% is barely. If you ignore the momentary donzuki, it can go up to about 200%. For all the files below, the log level for the fluid mix was 200% and that for each mix was 150%.
kanon_f_gogo.MP3 and kanon_e_gogo.MP3 are generated from gogo.dll (version 3.13a) using the Timidity ++ function. Bitrate is CBR 128kbps, 44100Hz sample rate, no emphasis, and set stereo (for comparison, all of the following are recorded in CBR + set stereo + 44100Hz sample rate without emphasis). It seems that the psychoacoustic model, both gogo.dll and lame, cannot be disabled in recent versions (it seems that it cannot be disabled unless the build option is changed and recompiled). It doesn’t matter much, but when generated by default, the ID3 tag’s “Genre” element is “Anime”. I don’t know if it’s Timidity ++ or gogo.dll, but it’s a pretty good initial setup.
kanon_f_128.MP3 and kanon_e_128.MP3 come out of Life to lame (version 3.96.1) with encoding rate 2 (default) + psychoacoustic model + no preset. The numbers at the end are 192 and 256, respectively, because only the bit rate is increased with the same settings. If you use VBR, etc., you can improve the sound quality even with the same file size. It took a bit of effort, and is clearly better than the sound produced by gogo.dll. This doesn’t mean that gogo.dll’s performance is inferior to poor, but it’s just a matter of prioritizing processing speed with a bit of cost-performance blindness to bitrate and sound quality. Previously I posted a file encoded at 0 encoding rate (higher sound quality mode), but lame 3.96.1 has a bug and it seems noise can be added so I replaced the sample.
In my ears, I played it with Lilith version 0.991 and from Creative Sound Blaster PCI-128 (Gift: Thanks HGW) to Matsushita’s SA-AK15 (Gift: the very common so-called “stereo”, but was the maintenance bad? I wonder? if it was originally something like that, when i heard it with aiwa’s HP-X122 (1500 yen discounted headphones for actual sale) through a thing about “sounds better than a 30,000 yen class radio cassette player” , the difference between lossless compression and 128 kbps licks is almost imperceptible even at high volume (at best, I feel like “the room I’m playing in has changed” – it’s also a placebo effect, so it’s as a bug). It’s a different story if you listen to it in front of you with speakers, but actually you rarely do it (or rather, it’s a nuisance to your neighborhood). So, I put the number 192 kbps at the beginning, but 128 kbps (160 kbps in the best case) is enough to listen in my room tion (well in my case it is too poor and the playback environment is poor. It’s very crazy so it might not be very useful).
Regarding the bit rate, 192 kbps is sufficient for MP3.
A raw Wave file (44100Hz x 16bit x 2ch = 1411kbps CD sound quality) occupies approximately 42MB in 5 minutes, less than 5MB at 128kbps, more than 7MB at 192kbps, less than 10MB at 256kbps, but it depends on the user or if you really want to stick to the reproducibility of the treble range. If you want to welcome sound processing, you can also use lossless compression. With lossless compression, the more striking the sound (higher SPL and thicker highs), the harder it is to compress, but even with lossy compression, such sound tends to get distorted unless it has a high bit rate. The expression of sound quality is strongly influenced by the playback environment, and indeed if you play it at a small volume with a sloppy device, even 128 kbps is indistinguishable from lossy compression.
I also prepared a sample file (including the file explained in the next point). kanon_f_200.ape recorded the aforementioned fluent mix sound with a 200% Timidity ++ amp, and kanon_e_150.ape also recorded the aforementioned eaw-mix sound with a 150% Timidity ++ amp. Monkey’s Audio used version 3.99 and set the compression level to standard (high compression settings only slow down the process and don’t shrink much). The recording time of the fluid mix is approximately 3 seconds longer, probably reflecting the difference in the last sustain. The mixture of fluids is easier to gain sound pressure and can reach up to approximately 250%. Also, because the sound pressure is high, compression is not very effective. eaw-mix has noise-like ripples after 15 seconds (and 55 seconds), and 150% is barely. If you ignore the momentary donzuki, it can go up to about 200%. For all the files below, the log level for the fluid mix was 200% and that for each mix was 150%.
kanon_f_gogo.MP3 and kanon_e_gogo.MP3 are generated from gogo.dll (version 3.13a) using the Timidity ++ function. Bitrate is CBR 128kbps, 44100Hz sample rate, no emphasis, and set stereo (for comparison, all of the following are recorded in CBR + set stereo + 44100Hz sample rate without emphasis). It seems that the psychoacoustic model, both gogo.dll and lame, cannot be disabled in recent versions (it seems that it cannot be disabled unless the build option is changed and recompiled). It doesn’t matter much, but when generated by default, the ID3 tag’s “Genre” element is “Anime”. I don’t know if it’s Timidity ++ or gogo.dll, but it’s a pretty good initial setup.
kanon_f_128.MP3 and kanon_e_128.MP3 come out of Life to lame (version 3.96.1) with encoding rate 2 (default) + psychoacoustic model + no preset. The numbers at the end are 192 and 256, respectively, because only the bit rate is increased with the same settings. If you use VBR, etc., you can improve the sound quality even with the same file size. It took a bit of effort, and is clearly better than the sound produced by gogo.dll. This doesn’t mean that gogo.dll’s performance is inferior to poor, but it’s just a matter of prioritizing processing speed with a bit of cost-performance blindness to bitrate and sound quality. Previously I posted a file encoded at 0 encoding rate (higher sound quality mode), but lame 3.96.1 has a bug and it seems noise can be added so I replaced the sample.
In my ears, I played it with Lilith version 0.991 and from Creative Sound Blaster PCI-128 (Gift: Thanks HGW) to Matsushita’s SA-AK15 (Gift: the very common so-called “stereo”, but was the maintenance bad? I wonder? if it was originally something like that, when i heard it with aiwa’s HP-X122 (1500 yen discounted headphones for actual sale) through a thing about “sounds better than a 30,000 yen class radio cassette player” , the difference between lossless compression and 128 kbps licks is almost imperceptible even at high volume (at best, I feel like “the room I’m playing in has changed” – it’s also a placebo effect, so it’s as a bug). It’s a different story if you listen to it in front of you with speakers, but actually you rarely do it (or rather, it’s a nuisance to your neighborhood). So, I put the number 192 kbps at the beginning, but 128 kbps (160 kbps in the best case) is enough to listen in my room tion (well in my case it is too poor and the playback environment is poor. It’s very crazy so it might not be very useful).
