To boldly go where no audiophile, or, perhaps just me, has gone before ...
As 'threatened' in a comment in abraxalito's blog I'm going to kick off posting and linking to some samples of sound generated by myself, at the moment examples of replay on my system, recorded using a very basic technique.
I'll probably regret it, but I'm going to put up a rough and ready trial run, to give a smell of what I'm aiming at. It's full of problems, but it gives some good indicators of what I'm dealing with, and hopefully some feedback will help me for future runs.
The system is my 'dud' HT all-in-one setup, heavily tweaked, and the recording device is a Fujifilm S4000 camera, which does HD video rather nicely, with mono 44.1/16 audio thrown in - that's handy. To maximise sound quality and minimise card storage used, hopefully, the lens was covered while recording, yes, no pics! Unfortunately, signal to noise is poor and the automatic volume control keeps falling over its feet, but you can't get everything for that sort of money :) ...
But it does do some things quite well, so it's worth the effort. The album used for this trial is Helene Grimaud, Chopin, etc, from this album: [URL]https://www.youtube.com/watch?v=upQ911nlCzw[/URL] - this gives an idea what the "raw" sound of the album is. Camera was set about 4 metres from the speakers, and I'm standing about a meter away from the camera, playing with the system remote. My recording is very faulty, at the beginning is me fiddling with a cover for the lens, and that dreary voice intruding is me announcing the volume setting: maximum setting is 40, and I announce that the following 'bit' is being replayed at 20, 30, 36, etc. 36 corresponds to a volume comparable to live, in room piano sound, and as is painfully apparent the AGC, auto gain circuit and mic in the camera struggles to cope with the intensity levels above about 25. This means that to properly capture high level sound levels I'll have to take the camera in another room, or shield the mic with some foam or attenuating material - suggestions welcome!
So, what this is intending to demonstrate is that the system can go up to decent levels without problems, and that I can drop the volume without losing "tonality". Of course, there are losses all the way in this recording path, so to minimise further degradation I've uploaded the recording as a mono WAV file. The audio gear is just in reasonable shape during this recording, not peak form, so that's another negative in this first round.
And it's here: [URL]https://www.mediafire.com/?8h90tn8ibf0ay79[/URL], sorry about the size but I wanted to start with a decent lump of a recording. This is all an experiment, so beee gentle ... :D
All brickbats, suggestions, etc, welcome!
Cheers,
Frank
I'll probably regret it, but I'm going to put up a rough and ready trial run, to give a smell of what I'm aiming at. It's full of problems, but it gives some good indicators of what I'm dealing with, and hopefully some feedback will help me for future runs.
The system is my 'dud' HT all-in-one setup, heavily tweaked, and the recording device is a Fujifilm S4000 camera, which does HD video rather nicely, with mono 44.1/16 audio thrown in - that's handy. To maximise sound quality and minimise card storage used, hopefully, the lens was covered while recording, yes, no pics! Unfortunately, signal to noise is poor and the automatic volume control keeps falling over its feet, but you can't get everything for that sort of money :) ...
But it does do some things quite well, so it's worth the effort. The album used for this trial is Helene Grimaud, Chopin, etc, from this album: [URL]https://www.youtube.com/watch?v=upQ911nlCzw[/URL] - this gives an idea what the "raw" sound of the album is. Camera was set about 4 metres from the speakers, and I'm standing about a meter away from the camera, playing with the system remote. My recording is very faulty, at the beginning is me fiddling with a cover for the lens, and that dreary voice intruding is me announcing the volume setting: maximum setting is 40, and I announce that the following 'bit' is being replayed at 20, 30, 36, etc. 36 corresponds to a volume comparable to live, in room piano sound, and as is painfully apparent the AGC, auto gain circuit and mic in the camera struggles to cope with the intensity levels above about 25. This means that to properly capture high level sound levels I'll have to take the camera in another room, or shield the mic with some foam or attenuating material - suggestions welcome!
So, what this is intending to demonstrate is that the system can go up to decent levels without problems, and that I can drop the volume without losing "tonality". Of course, there are losses all the way in this recording path, so to minimise further degradation I've uploaded the recording as a mono WAV file. The audio gear is just in reasonable shape during this recording, not peak form, so that's another negative in this first round.
And it's here: [URL]https://www.mediafire.com/?8h90tn8ibf0ay79[/URL], sorry about the size but I wanted to start with a decent lump of a recording. This is all an experiment, so beee gentle ... :D
All brickbats, suggestions, etc, welcome!
Cheers,
Frank
Total Comments 14
Comments
-
Yes, that first take is a bit rude but at least it's a start! I've been looking at perhaps putting up these efforts on something like YouTube to make it more convenient, less fiddling required by an interested party, but the bugbear that I've come across now many times rears its ugly head again: all the obvious sites use compression to some degree on the audio, and this knocks the quality badly -- at least on my machine, with my hearing ...
Have been playing with AAC encoding, as a supposed better "way", but no, still have the same problem. Even at the very highest quality level, meaning that the stream is locked at 256kbps, and comparing original WAV with that, using the Nero player, the difference is too dramatic. The Audacity program shows that I'm getting excellent encoding, using the iTunes code, the difference is very close to -60dB for the most part. But you wouldn't pick that from the hearing, over cheap PC speakers -- the WAV file has an ease and rounded tonality, it sounds like music; the compressed format version is flat, lacks sparkle, forces you to concentrate on the listening which means the fatigue level rapidily builds ... it's no contest.
Why is this so? My belief, as I've mentioned before, is that the processing required to decode the AAC format causes much higher levels of electrical interference to infect the DAC's workings, hence much greyer sound. What is really needed is a YouTube type site that has very straighforward streaming of WAV files to your desired player ...
Anyone know of one ...?
FrankPosted 26th December 2012 at 05:57 AM by fas42
Updated 26th December 2012 at 06:05 AM by fas42 -
Just a further thought about the Art of Listening: yes, the ear/brain can fool you, but in both directions! You can "make" two versions of similar playback sound the same in your head, just as you can make two identical ones different, by applying the grey matter inside your noggin!
The key to know what's going on, is to remain aware of, be sensitive to your mental state, specifically in terms of how relaxed you are, while listening. The better quality version of the recording will automatically be the benchmark for your brain, and when it hears a poorer version, it can "will" that to sound the same as the good one; but ... more mental exertion, stress, will occur while deciphering the lesser track. If you play around in a conscious way with this process you can literally become aware of this game your body is playing with yourself: listening to poorer sound, you recognise a level of tension in how you are listening, and so then you should, as an exercise, deliberately relax yourself, take the accelerator of how much focus, concentration you're putting into listening, and, if you're sensitive to it, immediately the sound will degrade, it will become dirtier, less pleasant, more artificial; so then you "tighten up" your hearing as the next step and, bingo, it immediately sounds much better again!
This audio is not an easy game, damn it all ...! :)
FrankPosted 26th December 2012 at 06:30 AM by fas42
Updated 27th December 2012 at 12:00 AM by fas42 -
Sorry, a bit slow with further samples, etc. Two problems: the recorder is not too brilliant, of course, there is a very narrow range where above that the AGC is too flaky, too intrusive, and below that the recording is overly noisy; and, an old problem re-emerges, the operation of the camera recorder affects the audio replay quality! Huhhh?! Well, this is still the current setup's major handicap, the internals are all exposed, there is no effective metal screening because the cover can't be re-attached easily and even if it were there are no guarantees that it would actually fully cure the "problem".
Which is, that RF interference degrades the SQ, even that being generated by a camera recorder! One would think that the level of the latter would be insignificant, but no, definitely lowers the "sparkle", life, of the sound. So I need to distance the recorder from the system, or experiment with screening of the recorder itself -- hmmm ... :scratch1:
FrankPosted 5th January 2013 at 10:22 PM by fas42
Updated 5th January 2013 at 10:27 PM by fas42 -
Still not happy ... if I get decent pickup on the recorder then I can still hear the AGC glitching ... so I'll investigate further whether this is normal, whether these sort of cameras are always struggling in this area.
In the meantime, have determined that I still have further issues with interference pickup in the house from mains power feeding various devices. So, either I disconnect the culprits, or put filters in the appropriate lines, or improve the shielding, the isolation of the audio gear. The last is obviously the best way, so another round of tweaking required ... oh dear, sigh ... :sigh:
Related to this, an excellent CD for ferreting out the finer points of bad behaviour, is a very rough transfer of Oscar Petersen: I Got Rhythm, from Past Perfect, 204379-203. The slightest abberation in the playback quality, and the piano tone becomes very hard to listen to, dulls and loses life severely at decent volume levels.
FrankPosted 7th January 2013 at 12:35 AM by fas42
Updated 7th January 2013 at 12:37 AM by fas42 -
Finally got a short clip of something to illustrate some points, that I'm reasonably happy with. This is off a Big Band collection CD, all tracks from the early '30s. The snippet is of Fletcher Henderson and his Orchestra, recorded in 1933.
This is running at just below maximum gain setting for the system, and was recorded on the camera recorder at the other end of the house, about 25 metres away. There's a clear passage linking the 2 areas, going around a couple of corners.
Hence the automatic gain of the camera is at maximum and it's quite noisy. But, the point I'm attempting to illustrate is that the tonality and sense of performance is quite nicely conveyed from the other end of the house, on a recording that could never be rated as being of premium quality!
Link is [URL="https://www.mediafire.com/download.php?83cjhcam9sekmb3"]BigBand-FletcherHenderson.wav[/URL], much smaller file this time.
FrankPosted 14th January 2013 at 01:40 AM by fas42 -
I mentioned White Limo, by Foo Fighters, as a particularly taxing track in which to resolve the musical lines, in one of the threads. Finally cracked it on the PC, after upsampling to DSD128, and running the track on repeat for a number of hours. So, another excellent stress test track to add to the arsenal ...
FrankPosted 16th January 2013 at 12:38 PM by fas42 -
What makes White Limo so hard to get "right" is that the drummer is using a ride cymbal, or equivalent, through most of the song. This is a similar situation with many Status Quo tracks, except that in White Limo what's added is the modern style of studio processing to give a very intense, in your face sound ...
So, on a normal system the response from some people could be, What ride cymbal? With improvements, or headphones, this instrument can be picked up, but then the goal is for the tone is be retrieved cleanly as volume increases, through speakers. Typically, the sound of the cymbal collapses back into the "wall of sound" as the system starts to stress at higher dBs, but this is what has to be worked on to avoid happening.
The final objective is for that ride cymbal to sit completely clear of the mix, in its own space, at maximum listening levels. Once there, the structure of the rest of the mix makes sense, and the song "settles down", becomes relatively straightforward to listen to ...
FrankPosted 16th January 2013 at 10:43 PM by fas42 -
On a forum I previously frequented there is a discussion of what "micro" and "macro" dynamics are. Someone posted a YouTube video, [URL="https://www.youtube.com/watch?v=3Gmex_4hreQ"]The Loudness War - YouTube[/URL], as an explanatory clip - this is really about the "bad" effects of studio compression, but is useful here in another context ...
The uncompressed version sounds good just on PC speakers, at a low volume. But if you played that on many "hifis", at a realistic volume, then people's complaints about system inadequacies make sense: if the intro runs at fully live level then the drum impacts may be too much, the amp will not handle them. Personally, I would expect a decent system to do this effortlessly, this is a decent recording which should be good to listen to at any volume.
The heavily compressed variation is in White Limo territory, the latter is squashed so that there are no shades of grey. Played soft it can sound like a tinny transistor radio, played loud it's just the same radio badly overloading. It's much harder to refine a system to enable the remaining dynamics in that track to make subjective sense.
The ideal or goal system is one that plays the extremely dynamic version at all volumes with no change of tonality, meaning that "micro" dynamics, low detail detail is audible and makes total musical sense, and "macro" dynamics, the big punchy, peak sounds are not compromised. And, as the second part of the package, to also be able to play the heavily compressed version well enough for it to still be musically satisfying, the detail is extracted well enough that the brain effectively recreates the dynamics to some degree ...
FrankPosted 22nd January 2013 at 11:24 PM by fas42
Updated 22nd January 2013 at 11:29 PM by fas42 -
Another bit of fun, I'm trying Dropbox at the moment, and this is another extremely unprofessional recording, from the time I was playing with this idea earlier in the year. Lots of noise from me fiddling with a cover to the camera lens, minimising the processing taking place in the camera while recording, plus you can hear the AGC of the camera still glitching from overload, even though the recording is done halfway down the house.
Don't ask me what the tracks are, I think it was a CD from the library, big band with singers - anyway, it gives a reasonable idea of what I'm after from music playback in the home ...
The link is [url]https://www.dropbox.com/s/8j3czfqgznb63v6/DropboxTest01.wav[/url]Posted 7th September 2013 at 03:57 AM by fas42
Updated 28th September 2013 at 02:21 AM by fas42 -
Quote:Have been playing with AAC encoding, as a supposed better "way", but no, still have the same problem. Even at the very highest quality level, meaning that the stream is locked at 256kbps, and comparing original WAV with that, using the Nero player, the difference is too dramatic.
Posted 13th September 2013 at 12:04 AM by benb -
Okay, try going a little bit further. I've mentioned an excellent YouTube video of a triple drummer effort, at a Buddy Rich memorial concert - [URL="https://www.youtube.com/watch?v=aFtxlEZhUAY"]390 Moments of Zildjian - 1989 Buddy Rich Memorial Concert - Louie, Dennis, Gregg - YouTube[/URL].
Plenty of material with rich - no pun intended! - harmonic content, good for testing systems. What I've done is grab the first minute or so of the audio from the video, and play it over the PC system, using the tiny Harmon Kardon computer speakers, running them essentially at maximum volume - this was possible because the audio track has very little level compression applied, and the speakers weren't noticeably strained by the exercise.
Now, I slightly 'cheated' - because there are high levels of treble in the track, and I found that my system produces better treble when the track is upsampled this is what the playback used - the WAV file upsampled to 352.8kHz. It makes a difference for me, :)!
To record the sound I used the camera per the first post here, in the largish spare bedroom where I do my computer fiddling. The sound levels would be too much for the camera AGC, so to overcome this I placed the recorder 4 metres from the speakers, but facing in the wrong direction - huuhhh!?! Well, it worked on first try, the sound picks up a lot of room bounce from the glass doors at the end, though.
The track played is this, mono and sampling rate of 44.1kHz, derived from the YouTube video at the highest quality setting: [url]https://www.dropbox.com/s/6g9cy2e6n1i0r94/ZildjianConcert-orig.wav[/url]
and this is what the camera picked up: [url]https://www.dropbox.com/s/hqt7djimlu52473/ZildjianConcert-rec.wav[/url]
The only fiddling was that the recorded version was dropped by 2dB to get a reasonable match, no other EQing or whatever was done ... . Oh, there's also a fade out, and fade back in about halfway through, done with the good ol' speaker volume control, as a sort of marker.
Well, it's obviously not a perfect match, :D, the tonal balance is way out on the 'copy' - will need the treble cut strongly to subjectively better align - and that treble is nowhere as good as it should be, obviously quite inferior! However, the way the recorder was positioned, picking up lots of room echo, and the quality of the playback system and camera recorder needs to be taken into account ...
So, what's the point of this exercise? Firstly, to show that the PC plus speakers does a quite reasonable job of sound reproduction, and secondly to move a step closer to fully closing the loop: have a recording of the playback of an original, and so be able to do some meaningful comparisons ...
Next, will attempt to improve the capture of that audio by acoustically attenuating the microphone pickup, so that it picks up much more of the direct sound ...Posted 26th September 2013 at 07:52 AM by fas42
Updated 28th September 2013 at 02:23 AM by fas42 -
On second look, listen, of the copy version not happy at all - just too far away in quality. Major part of the problem is that the AGC of the camera is too gung-ho, so heavy acoustic shielding will need to happen, to try and force the camera to work constantly at maximum sensitivity.
To get a bit better balance I matched the level of the background voices, by ear, at about 12 secs in, between original and copy. This meant an attenuation of 8dB, rather than 2, and it sounds more reasonable comparing the two.
So, added a 2nd version of the recording, [url]https://www.dropbox.com/s/c86zz1199ioeypn/ZildjianConcert-rec%2C-8dB.wav[/url][URL="https://dl-web.dropbox.com/get/Public/ZildjianConcert-rec%2C-8dB.wav?w=AABuJehW2J1SK6q2hkw8zq1b2HPOk-Kos-hG-dER_niVQw"][/URL], the only difference being that this has an extra 6dB of attenuation ...
The real progress will be made when I get the mic to pick up the direct sound, which means heavy acoustic attenuation. And then I will calibrate the loop to some degree, work out the equalisation to get the frequency response to match reasonably well.Posted 27th September 2013 at 06:11 AM by fas42
Updated 28th September 2013 at 02:23 AM by fas42 -
Apologies to anyone who was trying to retrieve the files in the last couple of posts, - %!&#*(@ Dropbox, which is supposed to be so easy to use, isn't - doing bizarre things for me, I'll eventually get it, I guess ... :D
Posted 27th September 2013 at 07:08 AM by fas42 -
Hilarious ... I just did another round of 'fighting' the Dropbox approach, and got myself in a mess again. The trouble with these websites, and software, which are 'so easy to use' (!!), that unless you do things exactly the way they expect the typical person to do things they become a nightmare - you go round and round in circles, fighting a will o' the wisp. This is where a little bit of knowledge is dangerous, one 'expects' things to operate a certain way, but they don't -- and confusion reigns.
Still not 100% happy that those files will behave themselves - but I've got my fingers crossed ... :D
Edit: should have Googled "Dropbox nightmare" before starting - got my head around this beast better ...Posted 28th September 2013 at 02:34 AM by fas42
Updated 28th September 2013 at 03:10 AM by fas42