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: LAME bit-depth and foobar2000 custom mode (Read 7232 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

LAME bit-depth and foobar2000 custom mode

I'm doing a custom setting in foobar2000 with LAME and it has a field called "Highest BPS mode supported" with the values: 16-bit, 24-bit, 32-bit and "32-bit float". Which one should I choose? Since I've read on several forums that the MP3 format "has no bit-depth", or that LAME simply ignores it. However, foobar warns that choosing an incorrect value could cause problems. Which should I choose? Or would any value be ignored by LAME? Regards.

Re: LAME bit-depth and foobar2000 custom mode

Reply #1 –
Hello Radagast.

You are correct, MP3 has no bit depth.
What you are talking about here is input format - stuff that you are feeding encoder with.
You can use 32-bit float with LAME.

Some encoders like Fraunhofer only accept 16/24-bit and if my memory serves me right some very old version of LAME don't like 32-bit float too.
gold plated toslink fan

Re: LAME bit-depth and foobar2000 custom mode

Reply #2 –
Quote from: Radagast on
I'm doing a custom setting in foobar2000 with LAME and it has a field called "Highest BPS mode supported" with the values: 16-bit, 24-bit, 32-bit and "32-bit float". Which one should I choose? Since I've read on several forums that the MP3 format "has no bit-depth", or that LAME simply ignores it. However, foobar warns that choosing an incorrect value could cause problems. Which should I choose? Or would any value be ignored by LAME? Regards.

I would select 16bps for everything because it would be guarantee that no problem can occur even if I don't know the highest one supported and the difference won't be audible at all. Even 8bps is totally transparent if the playing volume is a little bit low compared to the ambient noise, and 12bps is totally transparent in most listening conditions.

Re: LAME bit-depth and foobar2000 custom mode

Reply #3 –
Quote from: Klymins on
Quote from: Radagast on
I'm doing a custom setting in foobar2000 with LAME and it has a field called "Highest BPS mode supported" with the values: 16-bit, 24-bit, 32-bit and "32-bit float". Which one should I choose? Since I've read on several forums that the MP3 format "has no bit-depth", or that LAME simply ignores it. However, foobar warns that choosing an incorrect value could cause problems. Which should I choose? Or would any value be ignored by LAME? Regards.

I would select 16bps for everything because it would be guarantee that no problem can occur even if I don't know the highest one supported and the difference won't be audible at all. Even 8bps is totally transparent if the playing volume is a little bit low compared to the ambient noise, and 12bps is totally transparent in most listening conditions.
I think you are somewhat confused. He's asking about input bit depth, not bits per second.

Re: LAME bit-depth and foobar2000 custom mode

Reply #4 –
He isn't confused, 'bps' here means bits per sample. But the recommendation to intentionally limit the input to low bit depth makes no sense. While it makes no difference when encoding from unaltered CD source, it makes all the difference if encoding after DSPs.

Klymins can for example test what kind of MP3 he gets when encoding 20 dB volume boosted and 20 dB volume reduced MP3s with float input vs 16-bit input, just to give an example of the benefits of floating point.

Re: LAME bit-depth and foobar2000 custom mode

Reply #5 –
I guess that's me who is confused then!!  :-[  ;)

Re: LAME bit-depth and foobar2000 custom mode

Reply #6 –
Quote from: Case on
He isn't confused, 'bps' here means bits per sample. But the recommendation to intentionally limit the input to low bit depth makes no sense. While it makes no difference when encoding from unaltered CD source, it makes all the difference if encoding after DSPs.

Klymins can for example test what kind of MP3 he gets when encoding 20 dB volume boosted and 20 dB volume reduced MP3s with float input vs 16-bit input, just to give an example of the benefits of floating point.

I encoded 30db reduced Melodi 1 (from Supper Zill) with both 16bps and 32bps float using Lame (as FhG ones don't support 32bps float) and I didn't hear any difference. Also I noticed that I made a horrible typo on my previous post, sorry for that.

Re: LAME bit-depth and foobar2000 custom mode

Reply #7 –
Just leave on "32-bit float" as LAME works natively on that bitdepth.

Re: LAME bit-depth and foobar2000 custom mode

Reply #8 –
To the OP: as has been said, you should leave the input bitdepth at the default 32-bit float. The floating point format never clips so you can safely use any source or DSPs without fear of encoding nasty clipping artifacts. And it is highly precise even for super quiet tones - something that the other bit depths would truncate into background noise can still be processed with float.

Quote from: Klymins on
I encoded 30db reduced Melodi 1 (from Supper Zill) with both 16bps and 32bps float using Lame (as FhG ones don't support 32bps float) and I didn't hear any difference. Also I noticed that I made a horrible typo on my previous post, sorry for that.
I don't have Melodi 1, but Melodi 6 you have shared is super low fidelity and noisy (8 bits per sample, 8 kHz). It indeed doesn't seem to suffer much even with a little added noise. If my math is right 30 dB reduction still leaves 6 bits resolution left, that's just 2 bits less than initially.
Floating point benefit is easier to hear if you boost the loudness as you will immediately notice the results of clipping. Or if you for example reduce the loudness by 100 dB. 16-bit would have nothing but silence, float can still hear and see everything.
Your -30 dB reduction is ABXable for example with Pink Floyd's "Speak to Me" that has a long quiet fadein.

Re: LAME bit-depth and foobar2000 custom mode

Reply #9 –
Quote from: Case on
Quote from: Klymins on
I encoded 30db reduced Melodi 1 (from Supper Zill) with both 16bps and 32bps float using Lame (as FhG ones don't support 32bps float) and I didn't hear any difference. Also I noticed that I made a horrible typo on my previous post, sorry for that.
I don't have Melodi 1, but Melodi 6 you have shared is super low fidelity and noisy (8 bits per sample, 8 kHz). It indeed doesn't seem to suffer much even with a little added noise. If my math is right 30 dB reduction still leaves 6 bits resolution left, that's just 2 bits less than initially.
Floating point benefit is easier to hear if you boost the loudness as you will immediately notice the results of clipping. Or if you for example reduce the loudness by 100 dB. 16-bit would have nothing but silence, float can still hear and see everything.
Your -30 dB reduction is ABXable for example with Pink Floyd's "Speak to Me" that has a long quiet fadein.

Even if it doesn't clip at the encoding process, it'll probably do in the decoding process so I don't see the benefit. And I don't care about the other benefits so I still prefer to use 16bps.

And yes, Melodi 1 is 8kHz 8bps too. If anyone wonders, Supper Zill can be downloaded from the Internet while I don't prefer to give a website.

Re: LAME bit-depth and foobar2000 custom mode

Reply #10 –
Decoding clipping is easy to prevent during playback. Default configuration of foobar2000 for example would prevent it but not optimally, as it would be handled by Windows mixer and could cause volume fluctuations. Encoded MP3s can be losslessly volume corrected so that they don't clip.

Going with forced 16-bit input offers absolutely no benefits. It makes absolutely no sense to prefer it, except if the meaning is to get worst possible encodes.

Re: LAME bit-depth and foobar2000 custom mode

Reply #11 –
Quote from: Case on
Decoding clipping is easy to prevent during playback. Default configuration of foobar2000 for example would prevent it but not optimally, as it would be handled by Windows mixer and could cause volume fluctuations. Encoded MP3s can be losslessly volume corrected so that they don't clip.

Going with forced 16-bit input offers absolutely no benefits. It makes absolutely no sense to prefer it, except if the meaning is to get worst possible encodes.

I know, but I still prefer 16bps. Thanks.

Re: LAME bit-depth and foobar2000 custom mode

Reply #12 –
Quote from: Klymins on
Quote from: Case on
Decoding clipping is easy to prevent during playback. Default configuration of foobar2000 for example would prevent it but not optimally, as it would be handled by Windows mixer and could cause volume fluctuations. Encoded MP3s can be losslessly volume corrected so that they don't clip.

Going with forced 16-bit input offers absolutely no benefits. It makes absolutely no sense to prefer it, except if the meaning is to get worst possible encodes.

I know, but I still prefer 16bps. Thanks.

You can prefer 16 bit if it works well enough for your specific case but it is not helpful to recommend it to other people who may require higher dynamic range or be more sensitive to clipping.  Better advice would be to point out that 32 bit makes the most sense, but in some cases 16 bit will work well enough that the difference is not noticable. 

Re: LAME bit-depth and foobar2000 custom mode

Reply #13 –
Quote from: saratoga on
Quote from: Klymins on
Quote from: Case on
Decoding clipping is easy to prevent during playback. Default configuration of foobar2000 for example would prevent it but not optimally, as it would be handled by Windows mixer and could cause volume fluctuations. Encoded MP3s can be losslessly volume corrected so that they don't clip.

Going with forced 16-bit input offers absolutely no benefits. It makes absolutely no sense to prefer it, except if the meaning is to get worst possible encodes.

I know, but I still prefer 16bps. Thanks.

You can prefer 16 bit if it works well enough for your specific case but it is not helpful to recommend it to other people who may require higher dynamic range or be more sensitive to clipping.  Better advice would be to point out that 32 bit makes the most sense, but in some cases 16 bit will work well enough that the difference is not noticable. 

You're right, but I didn't give an advice. My sentence starts with "I would". But I assumed there's no significant DSP processing when writing "won't be audible at all", this is my mistake.