Compressed audio limits - Theory and practice

You mean like mp3's?

blutarsky, go here...

more information than you can shake a stick at...

My understanding:

With the right file, a well-trained ear can hear differences between compressed and uncompressed at any bitrate. But note that there are two qualifiers in that sentence: "the right file" and "a well-trained ear."

"The right file" is almost never a music file (unless you listen to some very weird music). It helps to have lots of high-frequency transients that aren't masked by other sounds. And "training" isn't just listening a lot; it's actually studying the particular sonic quirks of a codec and learning to identify them, first at low bitrates, then higher ones.

For people who listen to commercial music releases and haven't made a hobby of testing codecs, 192 kbps seems to be the point at which telling compressed from uncompressed starts to get really hard. That varies somewhat with the individual (though it's safe to assume that most people who claim to be able to tell the difference haven't really put themselves to the test).

That's the best place for it AFAIK.

I tolerate "Country Time" lemonade and instant coffee on occasion. Serves the purpose, but not the same as the "real thing" IMO.

I say the same thing about my plastic (she prefers to be called polymeric) girlfriend.

Let's correct some terminology that Mr. Blutarsky is using incorrectly.

An analog signal is SAMPLED at a fixed sample rate. Typical rates used in digital audio are:

32 KHz, 44.1 KHz, 48 Khz, 96 KHz, and 192 KHz. Note that the correct extension is KHz, NOT Kb/s. We are talking about a sample rate, NOT a data (bit) rate.

Now, those samples are quantized at various levels. Typical levels are:

8 bit, 12 bit, 14 bit, 16 bit, 20 bit, and 24 bit.

Now, when you mulitply the sample rate by the quantization level you get the data (bit) rate in bits/sec. For CD you get:

44.1 KHz X 16 bits = 0.7056 Mb/s per channel or a total data (bit) rate of 1.4112 Mb/s for a stereo CD. This is basic linear pcm encoding.

Uncompressed 5.1 PCM as recorded on a Blu-Ray disc is:

48Khz x 16 bits = 0.768 Mb/s per channel or a total data (bit) rate of 4.608 Mb/s. This is also basic linear pcm encoding.

Sample the analog signal at a HIGHER sample rate (96 KHz or 192 KHz) or a greater quantization level (20 bits or 24 bits), and the total data (bit) rate goes even higher. It is because of the huge data rates that result from linear pcm encoding that the various DD and DTS (or other) compression schemes were developed to save space on whatever medium is being used to carry the data.

Finally, any compression, perceptual coding, or other manipulation used to reduce the size of the signal file is applied to the final linear pcm signal. To put it simply if you take the 5.1 linear pcm signal used in my example and encode it with Dolby True HD (or any other LOSSLESS codec) and then decode it, the resulting pcm signal is IDENTICAL to the original linear pcm signal you started with. On the other hand, if you take my 5.1 linear pcm example and encode it with basic Dolby Digital (or any other LOSSY codec) and then decode it, the resultant output is still linear pcm but it is NOT identical to the original linear pcm signal you started with. It is missing data dependant on the codec used to process the signal. The missing data may be high frequency data, low level data, a combination or whatever the codec choose to discard.

Now back to our regularly scheduled discussion.

bluesky636,
(maybe) you forgot 88.2KHz and 32bit in the list but...
BINGO, cool post with perfects explanations.

Thanks.

I just listed the ones I was most familiar with.

bluesky636,

You seem to understand the issues well, but in the other thread blutarsky said "44KHz, 256Kbps, AAC" by which he almost certainly meant that the underlying PCM sample rate is 44.1KHz and the AAC bitrate is 256kbps; this is correct usage. I don't see anywhere that he said anything wrong; perhaps I missed it. In any case, given that we're talking about a pretty good compression scheme I'm forced to call BS on your subsequent quote that "digital audio from a phone is highly compressed crap that sounds like crap no matter what". I'm with mcnarus here.

I never said that Mr. Blutarsky's comment about 256 Kb/s data rate was incorrect. In fact, if you read my response in the other thread I compared that rate to the CD rate of 1.4112 Mb/s and said that a data rate of 256 Kb/s is crap.

The reason for my post in this thread was to clarify Mr. Blutarsky's continued reference to 192 Kb/s which is in fact an incorrect usage.

If you enjoy listening to cell phone audio and think it sounds good, more power to ya.

How is that incorrect usage? MP3s and such can be 192 kbps. Before you start correcting people, you ought to get your facts straight.

And even if it were a mistake, why did it take you 12 paragraphs to correct it?


If you think compressed audio and "cell phone audio" are synonymous, that's something else you've got wrong.

Well, I guess I can count on you to correct me, can't I?

And if I am not mistaken, this entire discussion arose out of Mr. Blutarsky's original thread about why his cell phone audio sounded like it had "wow and flutter."

Enjoy your low bit-rate. highly compressed audio, regardless of its source.

Thats what I was thinking.

Are you saying its not possible to have a high enough capacity SD memory card in a cell phone to permit storage and playback of high quality, uncompressed audio?

If so, you need to take a look at more cell phones.

I don't think I ever said anything like that in any of my posts. But if what you say is true, if Mr. Blutarsky continues to hear "wow and flutter" in his cell phone music, maybe HE needs to look at more cell phones.

However, I will readily admit that in my experience and opinion, low bit rate, highly compressed audio sounds like crap to me. Of course, there still remain some people who think that Redbook CD audio is the unholy spawn of Satan. Everyone has an opinion and is entitled to give it. It's a little something called "free speech."

Info theory ramble.

One good thing about the advance of technology (data storage and transmission) is that the need for compression is decreasing. The human limits of musical aural comprehension and the data requirements to store it uncompressed in at least 24 bit samples is becoming possible in the smallest packages.

As storage becomes unlimited the data transmitted to reproduce a selection becomes smaller because if every bit of information ever generated in already on your device, you only need to send selection codes.

For an audio only device this might be possible in our lifetimes but for all generated data it might be a little longer.http://www2.sims.berkeley.edu/resear...ble_report.pdf

You missed the point. You're assuming that any music on a cell phone is low bit rate and highly compressed. I maintain that if you use a memory card with sufficient capacity you could have lossless music on it. And yet you go on about low quality audio and cell phone audio as the same.