Android Others Project Update information about my project.

[FamVanHa]

Now, there are test results comparing the audio quality between popular audio codecs (MP2, AACv2, MP3, OPUS, AAC-LC) with my version of the Vorbis codec enhanced with advanced compression algorithms.

The official project name is "Advanced Audio Compression Algorithms with the OGG Vorbis Codec" (Improved Vietnamese alphabet: "Thwật Twán Nén Âm Thăjŋ Nâŋ Kaw Vớj Codec OGG Vorbis")
[Original Vietnamese Alphabet :
Thuật toán nén âm thanh nâng cao với Codec OGG Vorbis].

Audio codec Comparision at 128kbps:


Edit 1 (T2, 23/12/2024):
My previous post about this project owner: https://gbatemp.net/threads/i-inven...o-compression-at-36khz-128kbps-stereo.664429/
And...64kbps×2 is different from plain 64kbps because 64kbps×2 is almost identical to 128kbps.

Edit 2 (T2, 23/12/2024): added a video demonstrating how the advanced audio compression algorithm I invented works :


Edit 3 (T4, 25/12/2024) :
I need your comments on the output audio quality when encoding with the advanced audio compression algorithm (the audio track from 0:00 to 0:31 in the attached file is compressed with the Vorbis codec combined with this algorithm).
Anyone have any comments?!

 

[kijetesantakalu042]

Cool?

 

[CoolMe]

Using 48khz sample rate for lossy formats with lower bitrates is pointless (to say the least). Just downsample to 44.1khz like normal, and oh, you'd save a few mbs of space, since this is considered for video game use, no?

 

[ZeroT21]

The audio formats don't matter as much when using generic audio hardware to listen to them, but when you use audiophile grade hardware then it might matter as much. Reason some of my favorite old songs sounds like crap after upgrading to audiophile grade setup

 

[kijetesantakalu042]

Where is your audio format compared?

 

[CoolMe]

The audio formats don't matter as much when using generic audio hardware to listen to them, but when you use audiophile grade hardware then it might matter as much. Reason some of my favorite old songs sounds like crap after upgrading to audiophile grade setup

64kbps just sounds like crap no matter the audio gear. 128kbps is slightly better, still not nowhere near being ideal for any serious music listening experience. Same with 320kbps. Serious people go for lossless for a reason. But again, the use case is a bit different here if it's for video games, hence the focus on the OGG format..

 

[FamVanHa]

64kbps just sounds like crap no matter the audio gear. 128kbps is slightly better, still not nowhere near being ideal for any serious music listening experience. Same with 320kbps. Serious people go for lossless for a reason. But again, the use case is a bit different here if it's for video games, hence the focus on the OGG format..

A small note for you, the coefficient 64 kbps×2 does not necessarily mean 64kbps. 64 kbps×2 here means that the file after encoding will have a playback speed 2 times slower than the original audio file. To be able to play it back normally, we need a software that has the function of ×2 times the playback speed, after accelerating ×2 times, you can play the file back normally. When accelerating ×2 times, it means that you have doubled the original bit rate of the file. For example: the original file has a sampling rate of 48KHz, 16bit, 2 channels, with a duration of 3 minutes 8 seconds. After compression & slowing down, the file you receive will have the parameters of 24KHz, 16bit, 2 channels, with a duration of 6 minutes 16 seconds, 64kbps. You double the playback speed, the sample rate and bit rate are now 48KHz and 128kbps.

Post automatically merged: Dec 23, 2024

Where is your audio format compared?

Compare Vorbis X, a New Audio Codec with other popular audio codecs such as MP2, AACv2, MP3, OPUS, AAC-LC.

Post automatically merged: Dec 23, 2024

Sử dụng tốc độ lấy mẫu 48khz cho các định dạng mất dữ liệu với tốc độ bit thấp hơn là vô nghĩa (ít nhất là như vậy). Chỉ cần hạ tần số lấy mẫu xuống 44,1khz như bình thường, và ôi, bạn sẽ tiết kiệm được vài mbs dung lượng, vì điều này được coi là dành cho mục đích sử dụng trong trò chơi điện tử, đúng không?

Using 48khz sample rate for lossy formats with lower bitrates is pointless (to say the least). Just downsample to 44.1khz like normal, and oh, you'd save a few mbs of space, since this is considered for video game use, no?

in my algorithm, 64kbps×2 is roughly equivalent to 128kbps, the difference here is that my algorithm slows down the audio by 2 times, from 16bit, 48KHz, 2 channels, 3:08 to 16bit, 24KHz, 6:16, 64kbps, then decodes or plays back at ×2 speed to restore the original speed. The "×2" in "64kbps×2" is the slow down and speed up factor (I abbreviate it as HS)

 

[bobmcjr]

You need to understand that most audio codecs are engineered to maintain the quality of audio frequencies that humans are the most sensitive to. Slowing down the audio by half completely shifts all frequencies around, and libvorbis is going to de-emphasize the wrong regions and sound worse as a result.

If you are seeing smaller file sizes with this procedure, it's because vorbis is fooled by the half speed audio file, and is crushing otherwise important frequencies as a result. vorbis is a perceptual audio coding algorithm, and by lying about what frequencies are where, it will apply the wrong compression level to each frequency region.

tl;dr slowing the audio down tricks the encoder into removing/crushing important audio data and erroneously keeping the less important > ~20kHz data.

 

[FamVanHa]

You need to understand that most audio codecs are engineered to maintain the quality of audio frequencies that humans are the most sensitive to. Slowing down the audio by half completely shifts all frequencies around, and libvorbis is going to de-emphasize the wrong regions and sound worse as a result.

If you are seeing smaller file sizes with this procedure, it's because vorbis is fooled by the half speed audio file, and is crushing otherwise important frequencies as a result. vorbis is a perceptual audio coding algorithm, and by lying about what frequencies are where, it will apply the wrong compression level to each frequency region.

tl;dr slowing the audio down tricks the encoder into removing/crushing important audio data and erroneously keeping the less important > ~20kHz data.

I find the OGG Vorbis file combined with the advanced audio compression algorithm to be lighter (2.87MB) than the MP3 file (3.00MB) because of the lack of cover art and metadata, which seems more reasonable, because the MP3 file goes through the compression process directly from the source flac file, copying some metadata information and the Album art, while the OGG Vorbis file combined with the algorithm has a step in the encoding process that accidentally "removes" the cover art and metadata from the destination file.

Post automatically merged: Dec 23, 2024

You need to understand that most audio codecs are engineered to maintain the quality of audio frequencies that humans are the most sensitive to. Slowing down the audio by half completely shifts all frequencies around, and libvorbis is going to de-emphasize the wrong regions and sound worse as a result.

If you are seeing smaller file sizes with this procedure, it's because vorbis is fooled by the half speed audio file, and is crushing otherwise important frequencies as a result. vorbis is a perceptual audio coding algorithm, and by lying about what frequencies are where, it will apply the wrong compression level to each frequency region.

tl;dr slowing the audio down tricks the encoder into removing/crushing important audio data and erroneously keeping the less important > ~20kHz data.

and because after compression, the received audio file has been slowed down by 2 times compared to the original audio file, so when playing back or decoding, we need to add another step to this process, which is to increase the playback/decoding speed by ×2 so that the audio file can play normally. For example:
Original audio file: 16bit, 48KHz, 2 channels, 3:08.

File after compression with algorithm: 16bit, 24KHz, 2 channels, 64kbps, 6:16.

Playback or decoding at 200% speed:
16bit, 48KHz, 128kbps, 3:08.

 

[inversalAutoDoc]

Listen. I think the work you're doing is neat, really. I love esoteric file formats and weird edge-case support and all that junk. But like, I'm subscribed to the user news. I get a notification every time someone posts a new thread. I'm probably not the only one.

This generally works well. Most project threads on GBATemp are self-contained. People make a new post on their own thread for updates and update/edit the OP with an overview of their progress on major updates, it's a good system.

I don't think this is a rule or anything, but a request from one forum user to another: please don't make new threads with vague titles like this. If you have to make a new thread for one reason or another, please give it a descriptive name. A title like "Update Information About My Project" could be about literally anything - this is the Forum Where People Post About Their Projects - and spreading out your work like this makes it all harder to find later.

 

[FamVanHa]

I need your comments on the output audio quality when encoding with the advanced audio compression algorithm (the audio track from 0:00 to 0:31 in the attached file is compressed with the Vorbis codec combined with this algorithm).
Anyone have any comments?!