why does bitrate reduction lead to loss of highs?

Posted on

Member Since: Jan 18, 2003

when you upload to myspace, i think it re-encodes your song and brings the bitrate down. you can always hear a reduction in the highs. why does this reduction to the highs happen, and why to 'real bands' seem to get around this?

[ Back to Top ]


Administrator
Since: Apr 03, 2002


May 23, 2007 07:33 am

Most any form of compression seems remove the extreme highs and lows first and move more toward center the more the compression is applied. MySpace likely does this to shrink file size in order to save bandwidth and disk space and increase performance, since many users have music start playing on page load (which makes me leave, personally). Why compression works like that I can't tell you...what I CAN tell you is that HRC does NOT recompress or reprocess uploads in any way, shape or form. :-)

edit0r
Member
Since: Aug 17, 2004


May 23, 2007 07:47 am

Common sense tells me:

Highs = smaller wavelength + more detailed = needs more samples to reproduce 1 second timeframe

Lows = Longer wavelength + less detailed = needs less samples to reproduce 1 second timeframe

Therefore the easiest way to cut file size is to use less samples (accuracy) in HF's.

'Real bands' have mixes focussed around mid frequencies. Also the louder it is, the more bits your using.

I am not a crook's head
Member
Since: Mar 14, 2003


May 23, 2007 11:11 am

I agree dB, HRC has the highest fidelity user music around. I usually use either 160 kbps or 192 kbps for my music, and I've always appreciated the fact that our player here does not "dumb down" the bitrate like the other music sites do.

As a point of interest, when I signed up for MySpace, one of the calling points was "no compression on your music!". Pfffft. Total Bullshit.

Member
Since: May 15, 2004


May 24, 2007 08:32 pm

Thanks there CS.. that explains a lot! No wonder same thing happens when encoding to a lower bit rate mp3 format. Thought it was all from low quality dithering.

Chief Cook and Bottle Washer
Member
Since: May 10, 2002


May 25, 2007 09:34 am

Gotta ditto CS,

Rule of thumb per sample rate has always been publisized as 2X highest frequency to be reproduced. That is where 44.1 came from, roughly 2X 20Khz top of hearing range avg. That holds a thread of truth. Most folks can hear the difference between 44.1 and 48. 96, I don't know, haven't done any yet.

Per comment on 'real' bands or If I may be so bold 'well funded bands', I'm sure they take the time and effort to create a mix and master for each media they will be using e.g. radio, myspace, etc. CS is right. Most radio play stuff has a big frown eq on it to better match the bandwidth of the media. And then theres wav vs mp3. Infered sound. Yikes!

Duke of Stewed Prunes
Member
Since: Jun 01, 2007


Jun 01, 2007 08:16 pm

One thing that may have something to do with what you're hearing with compression (and I'd be surprised if anybody would argue that high frequency information was preserved at lower encoding rates...) is that high frequency information - pitch in particular - is something our ears are, by default, a lot more sensitive and perceptive towards than low frequency information.

While I have no interest in lab trials to test out this theory, it would probably stand to reason that low freq sounds are altered to a similar (if not the same) degree, but we simply can't hear the difference as readily.

Member
Since: Jun 02, 2007


Jun 03, 2007 02:46 pm

Walt hit the Nyquist theorum right on! Good response. Bit rate deals with dynamic range and sample rate deals with frequency response. When recording at 44.1khz the highest frequency you can record accurately is 20,050hz. Anything above this may produce aliases or false frequencies that are above the Nyquist frequency and got converted.

Time Waster
Member
Since: Jan 12, 2006


Jun 03, 2007 03:56 pm

Recording at 44.1KHz, a 20KHz sine wave looks like a triangle wave. A triangle wave and a sine wave do not sound the same. The A/D can only collect 2 samples (if that, errors happen all the time), and 2 samples is hardly enough to approximate a sine wave. I call the sound "brittle". You can really hear the difference when you record at 88.2KHz or higher. Much smoother, cleaner highs...

Hold 'Em Czar
Member
Since: Dec 30, 2004


Jun 28, 2007 02:21 am

how did i get the above post ^^^ in my e-mail?

said it was from [email protected]

Czar of Midi
Administrator
Since: Apr 04, 2002


Jul 01, 2007 12:13 am

did you click the subscribe to thread button?

Related Forum Topics:



If you would like to participate in the forum discussions, feel free to register for your free membership.