Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: About the bitrates of MP1, MP2, and MP3 (Read 1657 times) previous topic - next topic
0 Members and 4 Guests are viewing this topic.

About the bitrates of MP1, MP2, and MP3

Hello. I've found that bitrates of MP1, MP2, and MP3 have some probable disadvantages.

1: These bitrates are kinda weird:
    1.1: In MPEG-1 Layer 1, the bitrates are just multiplies of 32kbps.
    1.2: In MPEG-1 Layer 2, 40kbps isn't supported but 56kbps is supported.
    1.3: In all layers of MPEG-1 and MPEG-2 Layer 1, bitrates lower than 32kbps isn't supported while they can be enough for some sounds (especially for MPEG-2, but they can in MPEG-1 too, for example at the end of a music).
    1.4: 144kbps is supported for all layers of MPEG-2.
    Probably there are more oddities but these are what I could find right now.
2: The bitrates of MPEG-2.5 are just some of the ones of MPEG-2.
3: Discontinuous transmission isn't supported (it could be by allocating just one bitrate value for leaving the frame empty).
4: The bitrates are not dependent on the sampling rate (they could be given as bits-per-sample values instead of kilobits-per-second), and this causes the frames to have weird lengths on 44100Hz, 22050Hz, and 11025Hz sampling rates.

Why are these the case?

Re: About the bitrates of MP1, MP2, and MP3

Reply #1
You do realize that MPx formats were designed in the early 90s?
Some sacrifices had to be made so they could be used on devices from that era.
Why do you think we have newer and improved formats like MPC, Vorbis, AAC and Opus?
gold plated toslink fan

Re: About the bitrates of MP1, MP2, and MP3

Reply #2
SIGH! Here we go again... 
• The older, the lossier
• Listen to the music, not the media it's on.
• "União e reconstrução"

Re: About the bitrates of MP1, MP2, and MP3

Reply #3
Yes. Your findings are super interesting. Go on.

Re: About the bitrates of MP1, MP2, and MP3

Reply #4
You do realize that MPx formats were designed in the early 90s?
Some sacrifices had to be made so they could be used on devices from that era.
Why do you think we have newer and improved formats like MPC, Vorbis, AAC and Opus?

The conditions of my findings do not decrase complexity, and the last one even increases complexity (so the complexity would be lower if they did the opposite things).

The most important part of my question in different words: Does indicating the bitrate as a kilobits-per-second value instead of bits-per-sample in the stream have advantages?

Re: About the bitrates of MP1, MP2, and MP3

Reply #5
You are, of course, aware that only one of these things represents bitrate?
TAPE LOADING ERROR

Re: About the bitrates of MP1, MP2, and MP3

Reply #6
You are, of course, aware that only one of these things represents bitrate?

@itisljar Bits-per-sample represents bitrate too, as long as it's together with the sampling rate value.

Re: About the bitrates of MP1, MP2, and MP3

Reply #7
I mean, why the bitrate table wasn't for example something like this:

   0: 0bps
   1: 0.25bps
   2: 0.5bps
   3: 0.75bps
   4: 1bps
   5: 1.25bps
   6: 1.5bps
   7: 1.75bps
   8: 2bps
   9: 2.5bps
   10: 3bps
   11: 3.5bps
   12: 4bps
   13: 5bps
   14: freeformat
   15: invalid

?

Re: About the bitrates of MP1, MP2, and MP3

Reply #8
@itisljar Bits-per-sample represents bitrate too, as long as it's together with the sampling rate value.

Bits-per-sample is bit depth. You know, 4 bits, 8 bits, 16 bits, 20 bits, 24 bits, 32 bits. Each sample is represented by certain number of bits.
Are you sure you're not misreading bits-per-second?
TAPE LOADING ERROR

Re: About the bitrates of MP1, MP2, and MP3

Reply #9
@itisljar Bits-per-sample represents bitrate too, as long as it's together with the sampling rate value.

Bits-per-sample is bit depth. You know, 4 bits, 8 bits, 16 bits, 20 bits, 24 bits, 32 bits. Each sample is represented by certain number of bits.
Are you sure you're not misreading bits-per-second?

@itisljar No. Of course bits-per-sample does not represent bitrate on it's own, but it does when it's together with the sampling rate value. For example, 3bps at 44100Hz is equal to 132.3kbps.

Re: About the bitrates of MP1, MP2, and MP3

Reply #10
@itisljar Bits-per-sample represents bitrate too, as long as it's together with the sampling rate value.

Bits-per-sample is bit depth. You know, 4 bits, 8 bits, 16 bits, 20 bits, 24 bits, 32 bits. Each sample is represented by certain number of bits.
Are you sure you're not misreading bits-per-second?
Some lossy formats allow the user to set either a bitrate value or a bit-per-sample value. For example, Wavpack lossy allows these settings. If I set for example Wavpack to 400 kbps, it will use 400 kbps for a 44100 Hz sample rate (which is a high bitrate for CD material) but also 400 kbps for a 96000 Hz sample rate (which isn't high anymore, considering the file isn’t downsampled to 48000 Hz like most other lossy formats).
Setting the bit-per-sample value enables automatic bitrate adjustment according to the desired output sampling rate.
Wavpack Hybrid: one encoder for all scenarios
WavPack -c4.5hx6 (44100Hz & 48000Hz) ≈ 390 kbps + correction file
WavPack -c4hx6 (96000Hz) ≈ 768 kbps + correction file
WavPack -h (SACD & DSD) ≈ 2400 kbps at 2.8224 MHz

Re: About the bitrates of MP1, MP2, and MP3

Reply #11
"Bits per sample" has come to mean the number of bits representing each unpacked (pre-encoded) sample, not the average number of bits used at every sample point. Of course you can count the latter too: if you can compress CDDA down to 225 kbit/s, it corresponds to spending 5 bits per average sample, but "bits per sample" has by the development of the language acquired a different meaning.

I generally don't advocate over-moderation, but too much insistence to use phrases for their wrong meaning, will border on a TOS #10 violation. 

Example - hypothetical this time, but who knows next thread:
A "bitrate" of "50" does not mean a compression ratio of 50 percent. Of course, in a parallel wor(l)d, someone could have thought up that "bitrate" would be a nice phrase for bits per bit and it could have stuck.

Re: About the bitrates of MP1, MP2, and MP3

Reply #12
Some lossy formats allow the user to set either a bitrate value or a bit-per-sample value. For example, Wavpack lossy allows these settings. If I set for example Wavpack to 400 kbps, it will use 400 kbps for a 44100 Hz sample rate (which is a high bitrate for CD material) but also 400 kbps for a 96000 Hz sample rate (which isn't high anymore, considering the file isn’t downsampled to 48000 Hz like most other lossy formats).
Setting the bit-per-sample value enables automatic bitrate adjustment according to the desired output sampling rate.

Yes. And wavpack will encode both files with around 400 kbit/s, no matter what the sample rate (or original bit depth) is. Try it, I just did. It doesn't matter if it is too high bitrate for CDDA or too low for Hires, it will always be that.

By using one sample you are using one fixed point in time, and then you can't just change how many bits are assigned to that specific sample, because you will change the sound. Depending on sample rate and bitdepth, each sample of a song will have it's fixed value, no?
TAPE LOADING ERROR

Re: About the bitrates of MP1, MP2, and MP3

Reply #13
"Bits per sample" has come to mean the number of bits representing each unpacked (pre-encoded) sample, not the average number of bits used at every sample point. Of course you can count the latter too: if you can compress CDDA down to 225 kbit/s, it corresponds to spending 5 bits per average sample, but "bits per sample" has by the development of the language acquired a different meaning.

I generally don't advocate over-moderation, but too much insistence to use phrases for their wrong meaning, will border on a TOS #10 violation. 

Example - hypothetical this time, but who knows next thread:
A "bitrate" of "50" does not mean a compression ratio of 50 percent. Of course, in a parallel wor(l)d, someone could have thought up that "bitrate" would be a nice phrase for bits per bit and it could have stuck.

Thanks, but I don't know a word that I can use for the other meaning of "bits-per-sample".

Some lossy formats allow the user to set either a bitrate value or a bit-per-sample value. For example, Wavpack lossy allows these settings. If I set for example Wavpack to 400 kbps, it will use 400 kbps for a 44100 Hz sample rate (which is a high bitrate for CD material) but also 400 kbps for a 96000 Hz sample rate (which isn't high anymore, considering the file isn’t downsampled to 48000 Hz like most other lossy formats).
Setting the bit-per-sample value enables automatic bitrate adjustment according to the desired output sampling rate.

I think it's the opposite: setting the kilobits-per-second value enables automatic bits-per-sample adjustment according to the desired output sampling rate.


Anyway, I want to ask my question with different words (again): Why are the bitrates of 44100Hz same as the ones of 32000Hz and not the ones of 32000Hz multiplied by 1.378125 (44100/32000)?

Re: About the bitrates of MP1, MP2, and MP3

Reply #14
@itisljar No. Of course bits-per-sample does not represent bitrate on it's own, but it does when it's together with the sampling rate value. For example, 3bps at 44100Hz is equal to 132.3kbps.

So, by your calculations, CDDA audio has 32 bits per sample?

Please, use standardized language. It was enough to use "encoder colors", but this goes to another level.
TAPE LOADING ERROR

Re: About the bitrates of MP1, MP2, and MP3

Reply #15
@itisljar No. Of course bits-per-sample does not represent bitrate on it's own, but it does when it's together with the sampling rate value. For example, 3bps at 44100Hz is equal to 132.3kbps.

So, by your calculations, CDDA audio has 32 bits per sample?

Please, use standardized language. It was enough to use "encoder colors", but this goes to another level.

My equation was for mono, I forgot to mention, sorry.

Re: About the bitrates of MP1, MP2, and MP3

Reply #16
Yes. And wavpack will encode both files with around 400 kbit/s, no matter what the sample rate (or original bit depth) is. Try it, I just did. It doesn't matter if it is too high bitrate for CDDA or too low for Hires, it will always be that.
Try -b4 (as 4 bit per sample) with stereo files: it's around 350 kbps for 44100 kbps, the double for 88200 kbps, etc...
Wavpack Hybrid: one encoder for all scenarios
WavPack -c4.5hx6 (44100Hz & 48000Hz) ≈ 390 kbps + correction file
WavPack -c4hx6 (96000Hz) ≈ 768 kbps + correction file
WavPack -h (SACD & DSD) ≈ 2400 kbps at 2.8224 MHz

Re: About the bitrates of MP1, MP2, and MP3

Reply #17
Disadvantages compared to what?

how about bits per bitword ;)
And so, with digital, computer was put into place, and all the IT that came with it.

Re: About the bitrates of MP1, MP2, and MP3

Reply #18
Disadvantages compared to what?

1.1: This creates big gaps between the low bitrates.
1.2: This does not have a big disadvantage but is weird.
1.3: I explained this.
1.4: This is not from the standard order and looks unnecessary, limiting the available bitrates at the low side.
2: This limits the available bitrates of MPEG-2.5 to 8 while 14 could be used.
3: Discontinuous transmission can reduce the overall bitrate significantly in some situations.
4: This increases the complexity, and this is the one that I wonder the reason the most.

Re: About the bitrates of MP1, MP2, and MP3

Reply #19
1.1: This creates big gaps between the low bitrates.
1.2: This does not have a big disadvantage but is weird.
1.3: I explained this.
These codecs are not intended for ultra low bitrates!
The target bitrate for MP2 was 256 kbit/s, for MP3 128 kbit/s. Back in these days (the early 1990s) people compared the outcome with the original uncompressed sound and did not want to torture themselves.


Re: About the bitrates of MP1, MP2, and MP3

Reply #20
1.1: This creates big gaps between the low bitrates.
1.2: This does not have a big disadvantage but is weird.
1.3: I explained this.
These codecs are not intended for ultra low bitrates!
The target bitrate for MP2 was 256 kbit/s, for MP3 128 kbit/s. Back in these days (the early 1990s) people compared the outcome with the original uncompressed sound and did not want to torture themselves.

@Sunhillow I already said, but apparently I have to say again: Very low bitrates (like 16kbps for example) are completely transparent for some sounds (at high sampling rates too), like pure tones or humming, which can be encountered at for example the the silent parts of a music or a non-music recording.

Re: About the bitrates of MP1, MP2, and MP3

Reply #21
Try -b4 (as 4 bit per sample) with stereo files: it's around 350 kbps for 44100 kbps, the double for 88200 kbps, etc...

Yes, I am aware that 4-bit formats exist, like ADPCM. I am arguing against using that for showing bitrate instead standardized kbit/s or mbit/s.
TAPE LOADING ERROR

Re: About the bitrates of MP1, MP2, and MP3

Reply #22
I agree that bit per sample could be very confusing (with bitrate per second) with audio compression and video compression as well. But bit-per-sample is often used as measurement for still images instead of bitrate.
Example:
https://cloudinary.com/blog/jpeg-xl-and-the-pareto-front

It's also a valid way in audio, as for example David Bryant does without any debate recently:
https://hydrogenaud.io/index.php/topic,125991.0.html

I don't see any reason to argue with Klymins on that point.
Wavpack Hybrid: one encoder for all scenarios
WavPack -c4.5hx6 (44100Hz & 48000Hz) ≈ 390 kbps + correction file
WavPack -c4hx6 (96000Hz) ≈ 768 kbps + correction file
WavPack -h (SACD & DSD) ≈ 2400 kbps at 2.8224 MHz

Re: About the bitrates of MP1, MP2, and MP3

Reply #23
Try -b4 (as 4 bit per sample) with stereo files: it's around 350 kbps for 44100 kbps, the double for 88200 kbps, etc...

Yes, I am aware that 4-bit formats exist, like ADPCM. I am arguing against using that for showing bitrate instead standardized kbit/s or mbit/s.

I'm not saying that bitrate should be shown as bits-per-sample, I'm asking that does defining the available bitrates in a way that they don't change with the sampling rate have advantages (the opposite can decrase the complexity significantly)?

Re: About the bitrates of MP1, MP2, and MP3

Reply #24
I already said, but apparently I have to say again: Very low bitrates (like 16kbps for example) are completely transparent for some sounds (at high sampling rates too), like pure tones or humming, which can be encountered at for example the the silent parts of a music or a non-music recording.

    1.3: In all layers of MPEG-1 and MPEG-2 Layer 1, bitrates lower than 32kbps isn't supported while they can be enough for some sounds (especially for MPEG-2, but they can in MPEG-1 too, for example at the end of a music).
 

I don't remember any VBR Layer 1 or Layer 2 encoders. As said earlier, it's old technologies. Layer 3 was specifically standardized for efficiency. Layer 2 and 3 were not the best tools (especially with constant bitrate) for this purpose even if they probably perform well on specific cases. You just wondering why axes or chainsaw are not used to cut a sheet of paper. They do work, but if better tools exist, most people will use cutters or scissors and won't question why Stihl doesn't adapt their tools to cut your magazine.
Wavpack Hybrid: one encoder for all scenarios
WavPack -c4.5hx6 (44100Hz & 48000Hz) ≈ 390 kbps + correction file
WavPack -c4hx6 (96000Hz) ≈ 768 kbps + correction file
WavPack -h (SACD & DSD) ≈ 2400 kbps at 2.8224 MHz