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: 37800Hz on AAC (Read 9095 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

37800Hz on AAC

Hello. Does AAC support 37800Hz sample rate (and a half and a quarter of it)?

Re: 37800Hz on AAC

Reply #1 –
This says no.

There is one byte "code" for sample rate with 12 valid sample rates (0-B in hexadecimal or 0000-1011 binary).

Re: 37800Hz on AAC

Reply #2 –
Quote from: DVDdoug on
This says no.

There is one byte "code" for sample rate with 12 valid sample rates (0-B in hexadecimal or 0000-1011 binary).



Thanks. I wonder why they did not include 75600Hz, 37800Hz, 18900Hz, and 9450Hz. Also, I would prefer 6kHz, 5512.5Hz, and 4kHz instead of 96kHz, 88200Hz, and 64000Hz; I think using or supporting a sampling rate that is bigger than 48kHz is completely useless unless the goal is not listening it as a human, which does not have a high probability. The only use case I can think of is slowing down the sound, and in this case, I would prefer slowing down the sound before encoding.

By the way, it looked like a half-byte code to me.

Re: 37800Hz on AAC

Reply #3 –
Quote from: Klymins on
Thanks. I wonder why they did not include 75600Hz, 37800Hz, 18900Hz, and 9450Hz.

Non-harmonically related sampling rates are annoying to support in the encoder since the transform resolution will only be optimal for one sampling rate so it makes sense to limit the values as much as possible and instead resample the input/output if its too different.  We see this taken to the logical extreme with Opus where even 44.1KHz is not supported. 

Re: 37800Hz on AAC

Reply #4 –
MPEG-4 is quite a monster, and given how many "AAC"s there are ... who knows whether there is one configuration where the last item on the list at https://wiki.multimedia.cx/index.php/MPEG-4_Audio#Sampling_Frequencies can be used, and which technically can be called "AAC".
That list has a thirteenth rate of 7350 too.

37.8 can be Musepacked. Why not try?

Re: 37800Hz on AAC

Reply #5 –
Quote from: Klymins on
Thanks. I wonder why they did not include 75600Hz, 37800Hz, 18900Hz, and 9450Hz. Also, I would prefer 6kHz, 5512.5Hz, and 4kHz instead of 96kHz, 88200Hz, and 64000Hz;
Why making things unnecessarily complicated? You seem to be the only one around who loves ultra low bitrates and sampling frequencies

Quote from: Klymins on
I think using or supporting a sampling rate that is bigger than 48kHz is completely useless unless the goal is not listening it as a human, which does not have a high probability.
This is right for publishing music or for a transfer from analog tape or vinyl. While working on it in the studio higher resolutions are absolutely reasonable.

Re: 37800Hz on AAC

Reply #6 –
When I saw that 37800 sample rate it immediately made me think of PlayStation 1 CD-XA streams, which were usually encoded at that rate. It's 6/7 of 44100. I though you were going to ask for something like that.
Anyway, I believe Vorbis supports arbitrary sample rates, but I don't how badly that will affect quality.
If all you need is slowing down the audio, just resample after the slowdown or let the encoder do it if it has a resampler. Or does that interfere with your goal?

Re: 37800Hz on AAC

Reply #7 –
Quote from: radorn on
Anyway, I believe Vorbis supports arbitrary sample rates, but I don't how badly that will affect quality.

I can confirm it, Vorbis admits any sample rate below some limit (afaik 192khz), not sure about a lower limit.
Edit: I tested very low sample rates just right now. Lower limit seems to be 6 KHz.

Re: 37800Hz on AAC

Reply #8 –
The Vorbis spec says it was "intended for lower and higher sample rates (from 8kHz telephony to 192kHz digital masters)", but the audio_sample_rate field is a 32-bit unsigned integer with the restriction that "0" is forbidden.

Re: 37800Hz on AAC

Reply #9 –
Quote from: saratoga on
Quote from: Klymins on
Thanks. I wonder why they did not include 75600Hz, 37800Hz, 18900Hz, and 9450Hz.

Non-harmonically related sampling rates are annoying to support in the encoder since the transform resolution will only be optimal for one sampling rate so it makes sense to limit the values as much as possible and instead resample the input/output if its too different.  We see this taken to the logical extreme with Opus where even 44.1KHz is not supported. 

This is one of the reasons I hate Opus.

Quote from: Porcus on
37.8 can be Musepacked. Why not try?

Because it does not support anything below 32kHz.

Quote from: Porcus on
MPEG-4 is quite a monster, and given how many "AAC"s there are ... who knows whether there is one configuration where the last item on the list at https://wiki.multimedia.cx/index.php/MPEG-4_Audio#Sampling_Frequencies can be used, and which technically can be called "AAC".
That list has a thirteenth rate of 7350 too.

This is good, but did I see a "frequency is written explictly" in there? Doesn't this mean it can support all of the frequencies between a range @Porcus ?

Quote from: Sunhillow on
Quote from: Klymins on
I think using or supporting a sampling rate that is bigger than 48kHz is completely useless unless the goal is not listening it as a human, which does not have a high probability.
This is right for publishing music or for a transfer from analog tape or vinyl. While working on it in the studio higher resolutions are absolutely reasonable.

I think using a sample rate that is higher than 48kHz is definitely waste of bits, even in the studio, with the exceptions I've written before. Also, I think the same thing for bit depths higher than 16bps.

Re: 37800Hz on AAC

Reply #10 –
In MPEG-4, the sampling rate field has a code for arbitrary sampling rates to be specified (recall, MPEG-4 does allow for lossless formats too) - but that is not to say that all values are valid in all profiles. Whether there is an AAC profile where you can use that ... I don't know, really; from a source posted, it is apparent that there is some profile that does not.  https://wiki.multimedia.cx/index.php/Understanding_AAC#Packaging/Encapsulation_And_Setup_Data indicates there is some possibility somewhere, but that isn't to say that anyone has bothered to write an implementation.

Quote from: Klymins on
Quote from: Porcus on
37.8 can be Musepacked. Why not try?

Because it does not support anything below 32kHz.

So you make a post specifically on the topic 37800, and refuse to even try a codec that supports it - because it doesn't also support anything below 32?

Re: 37800Hz on AAC

Reply #11 –
Quote from: Porcus on
In MPEG-4, the sampling rate field has a code for arbitrary sampling rates to be specified (recall, MPEG-4 does allow for lossless formats too) - but that is not to say that all values are valid in all profiles. Whether there is an AAC profile where you can use that ... I don't know, really; from a source posted, it is apparent that there is some profile that does not.  https://wiki.multimedia.cx/index.php/Understanding_AAC#Packaging/Encapsulation_And_Setup_Data indicates there is some possibility somewhere, but that isn't to say that anyone has bothered to write an implementation.

Quote from: Klymins on
Quote from: Porcus on
37.8 can be Musepacked. Why not try?

Because it does not support anything below 32kHz.

So you make a post specifically on the topic 37800, and refuse to even try a codec that supports it - because it doesn't also support anything below 32?


Looks like I built a wrong sentence. I've tried Musepack, but I'm not planning to use it. Also, I didn't like the soft cutoff of it; I want a hard cutoff like MP3 does. This is also one of the reasons I hate Opus (it has a soft cutoff on fullband).

Re: 37800Hz on AAC

Reply #12 –
What is the purpose of these unusual sampling rates? They're out of any widely used standards.
TAPE LOADING ERROR

Re: 37800Hz on AAC

Reply #14 –
Quote from: Klymins on
Quote from: itisljar on
What is the purpose of these unusual sampling rates? They're out of any widely used standards.

For compatibility with other platforms and standards.
I thought you said it was for samplerate-based slowdowns...
What platforms are those anyway? I'm asking not only for 37800, but also the others you listed.
I'm curious now.

Re: 37800Hz on AAC

Reply #15 –
Quote from: Klymins on
For compatibility with other platforms and standards.
We are not children here. Please share which ones are you talking about with a bit more detail.
TAPE LOADING ERROR

Re: 37800Hz on AAC

Reply #17 –
Quote from: Klymins on
For example, the original PlayStation.

I thought PlayStation games used ADPCM type .XA files?  Are you developing a PlayStation game using AAC?  I simply use 44.1 KHz or maybe 32 KHz in that case.  Nothing is stopping from you using regular CD audio tracks on that platform which is 44.1 KHz.  Baring any limitations of the original PlayStation because I'm not familiar with it's limitations, I think using an ADPCM format would probably be a lot more appropriate choice since this is what it seems developers of that era often opted for.

Some screenshots of multi-platform game (The Grinch).
PlayStation is 37.8 KHz
Dreamcast & PC is 44.1 KHz

Re: 37800Hz on AAC

Reply #18 –
Have you tried ADPCM?
I recommend adpcm-xq by David Bryant, creator of WavPack.

Edit: Chibisteven was faster.
gold plated toslink fan

Re: 37800Hz on AAC

Reply #19 –
Quote from: Klymins on
Quote from: saratoga on
Quote from: Klymins on
Thanks. I wonder why they did not include 75600Hz, 37800Hz, 18900Hz, and 9450Hz.

Non-harmonically related sampling rates are annoying to support in the encoder since the transform resolution will only be optimal for one sampling rate so it makes sense to limit the values as much as possible and instead resample the input/output if its too different.  We see this taken to the logical extreme with Opus where even 44.1KHz is not supported. 

This is one of the reasons I hate Opus.

Will Opus let you set 37.8k as the output sampling rate?  If so it might actually be the best option.  Even if a codec supports an odd sampling rate like 37.8k, it is probably not going to compress as well as a standard sampling rate since it is unlikely any attention or tuning has been provided.  The advantage of Opus is that all sampling rates are treated optimally and then resampled to the target at output.  This is probably the best you can hope for aside from lossless.