Can someone tell me how can I disable the lowpass and highpass filters using a command?

Thank you!

Any particular reason why you want to disable filtering? Quality will suffer as a result unless you have a particular special problem case.

Cheers, Slipstreem.

You don't mention which encoder you're using, so I'm going to assume LAME.



Edit to add warning: Don't do this, unless it's for experimental reasons. Using this setting is stupid.

I somehow thought that if you disable some low frequencies, the bass may suffer. So I guess I was wrong?

Thank you, Slipstreem and Jensrex!

By default low frequencies are not filtered.

And the higher ones are those that produce the noise and undesired effects?

I simply meant that unless you include the --highpass switch, no highpass filter is applied.

Except perhaps when you use a preset and it has a filter in itself.

Do you have any reason to suspect this may be happening?

If so can you provide some evidence in the way of results from a double-blind test?

The presets will add a lowpass filter at the required frequency to optimise performance for a given bitrate or VBR -V value, however, none of them apply a highpass filter, AFAIK.

Cheers, Slipstreem.

Wait. In this case I mean the lowpass filter. And isn't e.g. in the V2 preset always 18.6 kHz?

Thereabouts, yes. Messing around with the filters via the command-line or any other means is almost guaranteed to lose you quality. Besides, you certainly won't lose any bass information by filtering out content above 18.6kHz.

Cheers, Slipstreem.

I know!



Do you know a web page or a site where there's an explanation why certain exact frequencies were taken as lowpass filters?

I'm afraid I don't, but I do know that thousands of hours of development work have gone into choosing them though. It's one of those areas of digital audio where I'm prepared to leave such things to people who understand lossy audio encoding much better than I do.

Personal ABX testing has proved to me that I can't distinguish between any CD source material and an encoding at VBR -V3, so I guess that high-frequency content above 17.9kHz doesn't matter to me personally in a complex waveform. Remember, we're dealing with music here, not test tones.

Cheers, Slipstreem.

You can modify the cutoff frequency by adding a --lowpass switch AFTER the -V 2 switch, but as Slipstream pointed out, you are more likely to do harm than good by doing this.

For current beta: it is 18.5 kHz for vbr-new and 18.6 for vbr-old.
For 3.96.1 it is 19.0 kHz (according to EncSpot).

So why did they lower it?

I'd suggest that you do some ABX testing. You're worrying over nothing.

Cheers, Slipstreem.

The lowpass setting is chosen to provide a "sweet spot" between leaving in high frequencies and having bits to spend on more audible frequencies. The developers are bettering that, in most listening environments, listeners are less likely to find an 18.5kHz lowpass audible than no lowpass and a reduction in the number of bits spent on other frequency bands. Considering that I only know of two samples I can ABX an 18kHz lowpass on (and none a 19kHz lowpass) this seems an excellent tradeoff for my purposes.

Fiddling with the lowpass setting is likely to degrade, rather than improve, quality. A set of tests (for example an ABC/HR setup) could tell you if changing this setting makes sense for you.

The biggest problem here I'm afraid to admit is... I don't know how to do an ABX test. Even with all these guides, I don't know if I should test a .wav or several .mp3's or... I'm just confused. :embarrassed:

No need to be embarrassed, but you really are worrying about this far too much in my opinion.

The easiest way to deal with the situation if you don't feel confident with the process of ABX testing is just to follow the Hydrogenaudio LAME WIKI. The vast majority of people can't tell an encoding at VBR -V2 from an original CD-quality source. It's a 'safe' setting to use for almost everybody. I settled for -V3 after extensive ABX testing and I honestly hear no difference personally despite having good ears for my age and good listening equipment.

Cheers, Slipstreem.

Well, you've all convinced me not to mess with the filters.

I just have to figure out how to do an ABX test, purely out of curiosity! I have to know how sensitive my ear is.

Don't be at all surprised if you try to abx some "killer samples" and fail. Being able to hear the kind of artifacts that the developers worry about requires some training.

On the other hand, there are those of us that don't want to train ourselves to hear those artefacts because then we would hear them whether we wanted to or not. Ignorance is bliss.

It sure is.

Does anyone know if there is a good freeware application that could analyze the frequency spectrum of an MP3 file?

The Analyse Spectrum tool in Audacity (a good free sound editor) will probably do what you want.

But you don't listen to music with your eyes, so why do you want to see it? It won't tell you anything that you can't hear.

Cheers, Slipstreem.

I know. I just wanted to see how certain other encoders cut certain frequencies off. For example, if someone rips something with iTunes at 256 kbps, I want to know if some frequencies were dropped. That kind of thing.

I would start by comparing Blade @ 192 CBR and Lame @ 192 CBR. If you can't tell the difference in a blind test then drop the bitrate to 160. Comparing your ABX results with a frequency response or spectral view of each sample should tell you all you need to know.

Well... with eig sample I can tell Blade@320 from Lame@320, but it isn't because of different lowpass value