hey all....
ok, I'm not trying to kick a dead topic...I swear..I'm just trying to give as much info as I've discovered on my lovely little player as I can.
so, here's what I did this time...I recorded it's actual audio output in to my computer for comparison. here are the technicals of the process:
the source file was from the self-released NIN album, The Slip. It was released under a Creative Commons license, so there is no copyright issues rising from my redistribution of the samples...the license allowed remixing and such. The original file was in FLAC format and was the "high-def" 96khz/24-bit version.
Using Adobe Audition I dithered this down to 96/16 using 96khz noise shaping which pushes the noise up above the audible range (25 - 30 khz). After dithering down to 96/16 was done, I then used the resampling feature with a good amount of pre/post filtering to arrive at the 44.1/16 files. I would of loved to have the player output straight 96/24, but doubting the DAC core in the RK2706 does 24-bit, I tried a 96/16 file. It plays...however it's expecting 44.1 khz and that's what it decodes at.
So, I loaded the downsampled WAV file on to my player and connected it to my M-Audio Transit USB...which is a really nice USB sound device (sometimes being an engineer helps you sort through the really good stuff and the overpriced crap). I maxed the volume on the player and recorded in to the PC at 44.1/24. I then took THAT file, doubled the sampling rate to 88.2 and then using the same noise-shaping profile I used to begin with, brought that file down to 16 bit and then sampled it back to 44.1.
Now, on the surface...the output from the player was pretty much exactly what I put in to it with the exception of being inverted (which could of been caused by the sound card) and the timing was off...it was about half a second longer and wouldn't sync up...but again...that could of been caused by the card and/or player...there's too many variables there to investigate that...however..as far as quality goes...it doesn't.
My particular player added virturally no noise that I could detect and the frequency response extended all the way just shy of 22khz, which is probably best explained by the filters on the analog side (there was a "band" between that and the extreme upper limit of the sampling rate which indicates so). The output on the sample level, despite being slightly out of sync...the waveforms were a 98% match, better than I had expected.
Over on the folder where I keep the pics of this player..I uploaded 4 audio samples. 1000000.flac, 1000000rk2706.flac, originalnoise.flac, and 2706noise.flac. It should be somewhat obvious that 1000000.flac is cut from the original file I copied to the player and the one with the rk2706 suffix is the same portion cut from the recorded sample.
The noise samples are just that...samples of very low level noise during the "silent" parts in the file. This noise barely indicated a blip on the spectrogram outputs, so I maximized it to bring any and all noise up to an audible range. The results were JUST ABOUT the same. There was the trailing end of the noise-shaping which got pushed down to between 21 and 22khz on the original, where as on the 2706 output, that portion is filtered..but the noise between the two files is just about the same. I would of done some real analysis of the THD and such, however, I don't have those tools available at the moment so I have to go on a pure "perceptional" quality rating. I measured the idle noise input of the M-Audio before starting and, just as expected, nothing.
Screenshot of the spectrogram/frequency domain of the amplified low-level noise from orig. file.
Spectrogram/Frequency domain of the same amplfied noise from the player's output.
The sample audio tracks are at
Index of /player
Re: yet another "what player is this"
