Lossy compression
so. i sorta like math. useful things at least.
so this didnt tire me. but im tired. head feels heavy.
anyway, should continue it. which i will... when im in the right mood.
developed this years ago. to re make.
so the third one is the max quality output in ableton live. it sounds like vinyl.
except, 60 mins is 11gb. well not much by todays standards.
if you record for 1 year, thats 97tb. or at VHS quality, 6tb. a 17x increase.
.
Lossy compression - the evolution in imagery while sound is still standing still at 1979 VHS standards and the 1990s 128kbps rule
info[redigera]
anyway, i wont update this. had an idea for another essay too. but wherever i post things like this online, i just get trolled.
text[redigera]
i found this question from 3 years ago:
https://www.reddit.com/r/mixingmastering/comments/r4cpf7/is_there_any_difference_in_sound_between_mp3_and/
anyone who posted opinions saying mp3 wasnt perfect, got trolled, downmodded and told to do the "double-blind" test.
the last post, which was also the most insightful, didnt though. i suppose this was because he was too late to the discussion.
for anyone wanting to do "double-blind" tests, try downgrading the bitrate in real-time. you will probably not notice the difference before going below 12-bit. your ears will eventually get fatigued though. same as listening to noisy recordings at high volume. the lower the bitrate, the more noise will be introduced. in ableton live, this is available from "redux legacy", just add the effect to a song, reset the downsample to 1.0, and switch around the bit reduction setting to get a feel.
the only logical bitrate to use, is 32-bit floating-point. why? if you output below this, your sound will be dithered twice. first by the DAW, then by various processes in your hardware and software.
another important point, is that mp3 is highly outdated. nowadays we have OPUS, which performs better than even AAC. back in in the 1990s, mp3 sounded horrible at 128kbps (11x compression ratio), with the cheap encoders that were then available for the layperson. but people and the industry, several decades later, are still implying 128kbps is flawless, eventhough we have gotten both an improvement in price and performances of DAWs, microphones, listening equipment and storage utilities.
to this day, streaming services and manufacturers insist on mostly using 128kbps for audio, eventhough for video it is up to 24mbps for 4k, and frame-rates have increased to 60FPS all the way to 120FPS in later standards, eventhough "most people" cant tell the difference compared to 30FPS. but the layperson will only be allowed to hear a stereo 16-bit 44.1khz encoded signal, that is still compressed with a 11x ratio.
soundcloud, on the other hand, uses only 64kbps in the free version of the website, and also employs both a limiter and compressor effect. this renders it impossible to listen to it, using headphones, without getting a headache, because of the high amount of noise introduced.
going all the way to 8k on youtube, this is finally "upgraded" to 152kbps OPUS for audio, but the video signal is a staggering 38mbps, meaning a ratio of 260:1 in video versus audio detail.
to really understand this though, we should talk about h.261 to h.266 codecs, and also redbook, dvd and blu-ray audio.
i will explain explain lossy audio, image and video compression.
"CD Quality" - finalized in 1979, for recording audio onto VHS tapes. this was then 1 year later chosen as the standard for CD audio (originally 650mb for data, and 74 mins for music). it has nothing to do with the amount of detail the human ear can hear. the bitrate is 1441½ kilobits per second, or 10 megabytes per minute, for easy reference. do note that data on CDs uses 13-bits per byte, because of error correction. so it is not possible to store the same amount of WAV audio in CD-ROMs for this reason.
DVD Audio - bitrate: 16-24, resolution 44.1-192khz, with a max of 96khz for full surround (5.1 channels). up to 9.6mbps losslessly compressed with MLP (the pure signal is up to 26mbps).
Blu-Ray Audio - surround was expanded to 7.1 channels. 2k res, 60fps, up to 4k in later versions, needing new hardware and discs.
h.261 - 1988. 40kbps-2mbps. made to be compatible with ISDN internet connections for live-streaming. 176x144-352x288 res, 1/16-1/4 of VHS.
h.262 - 1996. 704x576 res - full VHS support via interlacing. what is used on DVDs (5-17gb storage). "mp3".
h.263 - 2001. itunes was one of the first to start using this format. what is termed mp4 / m4a / aac, with vastly superior sound quality. mp3 encoders started copying many of its features, such as the open source LAME. one of the most important additions was variable bitrate, so a more complicated song (rock) would produce a higher bitrate than a less advanced (ambient). through recursive encoding it was now possible to create a specified bitrate even with VBR (quiet parts going down to 8kbps, and detailed ones up to 320kbps, while still for example maintaining 128kbps), also the ability to output a wanted final size. this is especially useful if your target customer has a certain max download speed, or to fit a movie on a CD.
h.264 - 2004. 1920x1080 (full-HD). what blu-ray discs use.
h.265 - 2013. 8192x4320, 0-120 variable FPS.
h.266 - 2020. 8-10 bits per channel. 360 degrees FOV. 2022 saw support for 12-16 bits (36-48 bit for LAB or 48-64 bit if an alpha channel [green screen] is also needed). this is needed to record HDR data for later processing, and high-end monitors (QLED) can also display it.
h.26* turns the sound, image and video content into mathematical equations.
with images and video, the image are chopped up into 8x8 pixel blocks. the higher the bitrate, the more advanced equation. even in the beginning, there was subsampling of the chroma part using the LAB colour mode, where 1/3 of the bandwidth is used for luminance, and 2/3 for colour. except the simplest subsampling already only uses 1/4 for the colour data (1:2:2) effectively making a 24-bit image the same as 2 channels (16-bit), down all the way to 1/16 (1:4:4) or 2-bit colour detail (10 bits in total).
similarily, with audio compression, a lot of bass and treble is removed, the lower bitrates you choose. another option available at low bitrates is to simply cut the frequencies at half of CD quality (max 10khz sounds) or more. the framerate for mp3 is 1152 (1024 for later), meaning, each second of audio (at 44.1khz) is broken down into 38 parts. each of which, at least later, can also variate in complexity.
for moving images, the standards have expanded on an ever increasing set of prediction algorithms.
JPG2K - never gained traction. the only difference is it blends together every block, which degrades the overall image quality.
the later JPG was improved to use many of JPG2Ks features, such as lossless compression (LZW) of the block data, and variable encoding. also progressive scan, which allowed better downloading of large files, as the image became more and more clear over time, instead of a stream.
written for https://www.quora.com/How-do-MP3-MP4-and-MP5-files-differ
i am appalled by the amount of misinformation here.
"is mushy, smeary, grainy, blurry and harsh compared to wav."
add stereo reduction info
av Tommy Mnemonic (ris och ros)
Aktivera autouppdatering av kommentar