Sangean HDT-1 Review

[audiophile.]

I knew something was fishy even with my first listen on speakers. AAC is easily disernable at 48k but 96k might been a slightly harder challenge. I know I could have got it right though....

 

[hipporadio]

I knew something was fishy even with my first listen on speakers. AAC is easily disernable at 48k but 96k might been a slightly harder challenge. I know I could have got it right though....

At one point in my life, I COULD HAVE discerned the difference at 96k—but those days have passed. I can still spot a 128k mp3—but that’s an easy one! Back in the 80s, I was “audiophile”—NO—“rabid audiophile”, but these days I sometimes find myself asking folks to repeat themselves in a noisy room! Today, I prefer to enjoy CONTENT over audio capacity—and I believe MOST radio listeners do also.

I know NOTHING about the mechanism he employed in his third (and latest) “demo”—I was uninterested in participating—because I had a “cautious question” about his second and a portion of his first. I don’t know if the files were available only as a live stream or they could be downloaded. I assume the latter was at least possible, and opens up the potential for a second level of trickery that could have allowed him to “have it either way”. Consider in fairness...If he NEVER produced a sample at 48k that was unconverted; his intentions, while misguided, may have been innocent. If ONLY as a stream, I can’t imagine conducting a fair evaluation on the speakers that accompany most computers!

Point is, Mike (or ANY of US) would not be in a credible position to offer a “controlled test”—unless of course you accept that we are the only ones at the controls. We have NO IDEA what can transpire in the studio of any man employed in “audio production”... I only know that even with my “average” audio resources in my own studio—I can easily mimic a “processed” off-air station on either band with the appropriate production parts.

 

[Mike Walker]

You're right Hippo. You have no way of knowing. It's a matter of trust. Having read my words here, you either consider me worth of trust, or not. It's your call.

I am merely trying to bring some actual AUDIO to our discussion of, and arguments about AUDIO. It takes time from my schedule, time that I could spend continuing my new John Lescroart book (The Suspect...great so far!), or getting actual work done. I'm trying to contribute something of value...something to actually LISTEN TO, to bring to the debate.

As for making the compressed files available, well...that would defeat the purpose of the tests...which is to discern differences by LISTENING. I see no reason not to make the compressed files available AFTER the tests are concluded, in the future. And if those here would consider that of value, then I'd be glad to.

What I wish others would do rather than bitching about files I post is POST YOUR OWN! It's a little work, but anything of value in life is. Don't just tell us how things sound, SHOW US! We're radio geeks, aren't we?

 

[Mike Walker]

Hippo I'd point out that before compressed digital audio can be listened to on our "analog ears", it must first be decompressed to linear pcm, then converted by digital to analog converters. It makes no difference if the conversion from compressed back to linear pcm takes place "on the fly" during playback, or in the form of an uncompressed pcm file...it's the same process. Digital to analog converters expect to receive linear pcm, and that's the form the audio must be in before being decoded. So your example that my work was "doubly compromised" (I think were your words), is doubly wrong. Conversion back to PCM WILL TAKE PLACE before conversion to analog. It's just a question of when that conversion takes place.

And linear pcm files of the same length (same number of samples) would have been exactly the same size whether they had been converted to aac+, mp3, or if they were silent (all zeroes). That's the inefficiency of pcm audio...it takes the same amount of data whether all sixteen (or 24) bits are exercised (loudest sound possible), or none of them are (no sound recorded). Either way, there are 44,100 (or 96,000...whatever) samples of 16 (or 24) bits taken each sample. Example below.

1111111111111111
0000000000000000

Two digital words...one representing silence, one representing the loudest sound possible to record. Both use the same amount of data.

 

[Tom Wells]

Mike, the link posted to the audio waveform sine wave-vs-sawtooth-vs-square wave dicussion supports my assertions.

Fourier-transform derivation must presume sinewave reconstruction.
Near the end of the cited article, is the information that overtones of many instruments have non-sinusoidal charactersitics.
This is what distinguishes individual voices.. of people and instruments... the balance of tones.

The conclusion of the article sums up that perhaps we should be sampling at much higher rates, even though we only hear to 20 kc, excuse me khz.

Don't get me wrong, CDs are good, but I consider the medium also useful for a hiss and click limiter in the same way
I used cassettes as a very effective click limiter and peak limiter for broadcast.

A 3-head deck with live-record monitoring is a fine poor man's peak limiter for a shortwave pirate station.

CDs are just like "soft focus" used in photography to hide tired old lines on faces in portraiture.

Will you be treating us to any of your Audio Fidelity discs in sampling, so the various rates and codecs can be compared for those
who wish to test their ears and upstream equipment?

I want to hear how nice that little rock follows the gutter in NC, and how the audio characteristics of your cartridge/arm sound.
Just like certain violins are magical sounding, analog can be bad to good. Show us your ability to do analog well.
The differences in encoding should be glaringly obvious, with a "near live" quality source.

 

[Mike Walker]

But you're forgetting something Tom. Overtones in music are harmonically related to the fundamental. If a fundamental tone is 20khz (or even 15khz...and no instruments produce musical notes with fundamentals this high!), then the overtones would be at least at 30khz, or 40khz. Which means they'd be beyond the limits of both hearing and 44.1khz audio. So an ACCURATE reproduction of this fundamental, minus it's overtone, would be a sine wave.

Take a complex waveform, and remove all harmonically related overtones. What's left? A SINE WAVE! So a sine wave IS an accurate reproduction of high frequency fundamentals minus their overtones.

Nothing screws with phase/time relationships like analog reproduction! Record a ten cycle 10khz burst digitally. Observe it on an oscilloscope. It starts precisely, and ends suddenly after exactly ten cycles. Do the same on a phonograph. The first thing that happens is that the stylus, suddenly forced into action, GROSSLY overshoots the first cycle (continuing to move, when it should have stopped and reversed course) resulting in a large spike. Then after finally almost settling down after a few cycles, the tone is removed...but the stylus continues to vibrate...producing ringing for many cycles thereafter. The resulting waveform looks absolutely nothing like what went in. And this is with simple sine waves. Imagine how it mangles complex waveforms like music!

 

[Mike Walker]

Well...here's a song I encoded for my own enjoyment from an old Nicolette Larson album (still in great condition, though the pressing is 30 years old).

It's in mp3, however. I have limited space on my website for huge, uncompressed files.

http://www.theproductionroom.net/nicolette.mp3

Unless I'm horribly mistaken, this recording was never released on cd (I've never seen it), despite the fact that it was a huge seller. A great argument for having a turntable, even if you listen primarily to your 'Pod!

 

[Tom Wells]

Yes, I admit the mass/acceleration limitations of the stylus and shank are the most critical in viynl source audio.
I fully agree that under/overshoot on transients is the first place viynl shows weakness.
Naturally, the lightest, smallest stylus and shank tracks the groove best.

Regarding the two-tone test, we did this back in '81 in lab, and were shocked that some people did, and some people did not hear a chord
when the two tones were greatly differing levels.

It sounds as though you have a better cartridge and stylus than I have.

The 192K sample rate is quite enough for me to make this determination.

Acoustics in listening evironments also play havoc with phase time relationships, and hearing music in a live environment often
incorporates similar degradation to what happens with typical analog loss.
Our ears accept some types of "data loss" as natural.
Rolling off of high frequencies as in old fashioned telephone response sounds more natural than a time-garbled low bitrate cellphone.
Muffled sound is as old as the hills, but time-chopped audio is still new thing to our ears.

If HD were using 192K, I would have no beef about the capabilities.
I have heard 96K sound wonderful and 48Ks that were acceptable for non-critical listening.
Maybe what I've been hearing on HD FM is the pre-emphasis not being removed for the HD audio.
I am still wondering why the violins on the WDAV clip had excess "zing".

Your encoding of analog and digital sources both surpass the audio quality of HD FM I have heard.

I am also considering how well the 192k sampling reproduces the tiny "pops" in the audio.
I will also have a listen to this with the headphones when I get home tonight.

Just as with digital photography, when enough data is involved the resolution becomes transparent.
I guess I'm saying I find the current chosen standard to be marginally acceptable in resolution.
I hope the future holds higher bitrates for streaming audio and HD FM.