Unoffical empeg BBS

Quick Links: Empeg FAQ | Software | RioCar.Org | Hijack | jEmplode | emphatic
Repairs: Repairs | Addons: Eutronix | Cases

Topic Options
#368349 - 16/02/2017 20:08 Empeg and Rio guys: question about 48khz sample rates on Empeg etc
tfabris
carpal tunnel

Registered: 20/12/1999
Posts: 30505
Loc: Seattle, WA
I did a mixing session for a video for Molly Lewis recently, and since we were taping for a video, at the request of the video guy, we recorded all the audio tracks at 48khz because that's what his video editor software expects to receive.

We mixed everything in Logic Pro X for the Mac, which has an option to bounce down the final mix to MP3. So I did so, and slapped that MP3 onto my Empeg Car Mark 2, to preview the mix during my commute. To my horror, it sounded distorted all the way through. It was faint and subtle distortion, but definitely audible, especially around the high notes in the lead vocal.

Oddly, the MP3 sounded fine when I played it back on any other gear such as other computers and such. No distortion. Later I looked more closely, and WinAmp's "file properties" box tells me that the MP3 was encoded at a 48khz sample rate: Logic didn't even offer us an option to convert the sample rate to 44.1khz, so it just did a straight 1:1 encode.

Remembering that the Empeg uses a DAC originally designed for car CD players, I'll bet that the DAC is hard-wired to accept only 44.1khz input, am I correct? I'm convinced that the 48khz sample rate of the MP3 is the culprit, and the reason I heard the distortion. The type of distortion I heard was consistent with what a playback engine would sound like when it did a poor conversion from 48 to 44.1 (i.e., it sounded like a direct "just drop the samples you don't need" conversion, without any noise shaping or dithering algorithm).

I want to convince my friend who is on the Logic Pro X team at Apple, that the lack of samplerate conversion options on their MP3 bouncedown feature is a bug. To do this, I will need to know that the Empeg Car is not the only hardware MP3 player that doesn't play 48khz MP3s well. And it occurs to me that the people who would know this would be the Empeg/Rio hardware guys, folks like Hugo and Patrick.

So my questions to you guys are:
- Is it true that the Empeg won't play 48khz MP3 files well, resulting in distorted playback?
- Is it true that the issue is that the DAC is a CD-player DAC which can only accept 44.1khz properly?
- Does anyone know of any other hardware-based MP3 players that will have the same problem?
- Can anyone think of other situations where a 48khz MP3 file will play back poorly on other kinds of equipment?

Thanks!
_________________________
Tony Fabris

Top
#368350 - 16/02/2017 20:19 Re: Empeg and Rio guys: question about 48khz sample rates on Empeg etc [Re: tfabris]
tfabris
carpal tunnel

Registered: 20/12/1999
Posts: 30505
Loc: Seattle, WA
(It would be particularly interesting if anyone with Apple hardware knowledge knew if any older Apple iPod might have this issue.)
_________________________
Tony Fabris

Top
#368351 - 16/02/2017 20:34 Re: Empeg and Rio guys: question about 48khz sample rates on Empeg etc [Re: tfabris]
tfabris
carpal tunnel

Registered: 20/12/1999
Posts: 30505
Loc: Seattle, WA
I think I know of one other situation. You know how some car CD players will also play discs full of MP3 files? I'll bet those players, at least in the olden days, probably also had 44.1khz DACs, and I'll bet they'd choke on 48khz files too.
_________________________
Tony Fabris

Top
#368352 - 16/02/2017 22:09 Re: Empeg and Rio guys: question about 48khz sample rates on Empeg etc [Re: tfabris]
mlord
carpal tunnel

Registered: 29/08/2000
Posts: 13443
Loc: Canada
You could use sox to convert it to any format/rate you like, and try that on the empeg to see if 44.1 is indeed better for it than 48.

Top
#368353 - 16/02/2017 22:34 Re: Empeg and Rio guys: question about 48khz sample rates on Empeg etc [Re: mlord]
tfabris
carpal tunnel

Registered: 20/12/1999
Posts: 30505
Loc: Seattle, WA
Certainly. I'm aware that, now that I know there's an issue, I can work around it by outputting to WAV or AIFF at the desired samplerate and encoding the MP3 from there using third party external software.

The real issue is that the MP3 bounce down feature in Logic is there for the sole purpose of allowing me to skip that external step in my production pipeline. If the feature is lacking, then it doesn't save me the step and so the feature is bugged.

What I really want to know is: is this really the issue I think it is, and if so, how widespread? If the empeg-car is the only MP3 player that can't handle these files, then maybe it's not worth their time to fix the issue.
_________________________
Tony Fabris

Top
#368354 - 16/02/2017 23:03 Re: Empeg and Rio guys: question about 48khz sample rates on Empeg etc [Re: tfabris]
mlord
carpal tunnel

Registered: 29/08/2000
Posts: 13443
Loc: Canada
Until you try a reconverted file on the empeg, it will be difficult to know what the problem is.

Perhaps the sampling frequency is actually fine, and instead the empeg is ignoring some kind of global gain/volume adjustment, leading to the distortion you hear. Dunno.

Top
#368355 - 16/02/2017 23:32 Re: Empeg and Rio guys: question about 48khz sample rates on Empeg etc [Re: tfabris]
tfabris
carpal tunnel

Registered: 20/12/1999
Posts: 30505
Loc: Seattle, WA
Ah, I see what you mean. As a diagnostic step, not as a work-around to the production pipeline. Gotcha. I'll try that.
_________________________
Tony Fabris

Top
#368356 - 17/02/2017 00:28 Re: Empeg and Rio guys: question about 48khz sample rates on Empeg etc [Re: tfabris]
tfabris
carpal tunnel

Registered: 20/12/1999
Posts: 30505
Loc: Seattle, WA
Blech. Took a while to get Sox working because it doesn't come with the necessary DLLs to run it, it gets a missing DLL error when you try to run it and you have to go hunting for some other DLLs (nothing mentioned on their dox about needing to do this).

Then the dox kind of suck in terms of being able to easily locate the quality settings. I can't tell how to make the thing do its MP3 encoding or its sample rate conversion at its highest quality settings.

However, even with all of the above suckiness....

I was able to figure out how to get it to convert the "bad" mp3 file to a new mp3 file like this:
Sox.exe inputfile.mp3 outputfile.mp3 rate 44.1k

That got it to samplerate convert the file to 44.1k at its default low-quality resampling method, and to re-encode it at its default low-quality 128k MP3 quality.

And the resulting file, played on the empeg, sounds significantly improved. Though it's still a crappy 128k MP3, and I can hear the crap due to that, there is none of the fuzzy distortion surrounding the vocals and instruments that the original file had.

Though the above involves a complete rework/resample/re-encode of the file, so I'm not certain it proves anything, the fact that the distortion is gone at 44.1k seems to me to be a good indicator that the samplerate was indeed the culprit.
_________________________
Tony Fabris

Top
#368357 - 17/02/2017 18:30 Re: Empeg and Rio guys: question about 48khz sample rates on Empeg etc [Re: tfabris]
mlord
carpal tunnel

Registered: 29/08/2000
Posts: 13443
Loc: Canada
Originally Posted By: tfabris
Though the above involves a complete rework/resample/re-encode of the file, so I'm not certain it proves anything, the fact that the distortion is gone at 44.1k seems to me to be a good indicator that the samplerate was indeed the culprit.


So now just repeat some of that minimally, except at 48KHz, and see what the empeg does with it.

Top
#368358 - 17/02/2017 19:16 Re: Empeg and Rio guys: question about 48khz sample rates on Empeg etc [Re: tfabris]
tfabris
carpal tunnel

Registered: 20/12/1999
Posts: 30505
Loc: Seattle, WA
Ah yeah, good idea. I can try that.

Before I do that, though, it'd be nice if I could find the command line option in Sox to increase the quality of the MP3 encoding when it re-encodes. Some of the artifacts that come up at its default crappy 128kbps encoding are similar to bad resampling noise, and so I have to listen closely to be able to tell the difference. I can't find it in their dox.

And their dox on the resampling options (rate -q|-l|-m|-h|-v) are super-unclear as to which ones get you the highest quality.

And finally the dash character they show in front of parameters in the doc file isn't a hyphen (even though it's a hyphen in real life), it's some other hyphen-like character (−), so when I search the docs for "-q" for example, it doesn't find any the instances of the q parameter in the docs, even when I'm staring right at one right there. Boy that's annoying.

Anyway, if you can tell from that thing how to control the MP3 encoding quality, let me know.

In the end, I think I've already got the answer to my first question: "- Is it true that the Empeg won't play 48khz MP3 files well, resulting in distorted playback?" - The answer is TRUE.

Now I am still wondering about the other questions:
- Is it true that the issue is that the DAC is a CD-player DAC which can only accept 44.1khz properly?
- Does anyone know of any other hardware-based MP3 players that will have the same problem?
- Can anyone think of other situations where a 48khz MP3 file will play back poorly on other kinds of equipment?

_________________________
Tony Fabris

Top
#368359 - 17/02/2017 20:20 Re: Empeg and Rio guys: question about 48khz sample rates on Empeg etc [Re: tfabris]
mlord
carpal tunnel

Registered: 29/08/2000
Posts: 13443
Loc: Canada
Not sure what form of documentation you are trying to look at there, but here I just do "man sox" and it pops up in gory detail. smile The "hyphens" you see are "-" minus signs, as with any other unix command.

Here's the info on that rate option you asked about. My first guess was that 'q' meant "quick and dirty", and l/m/h stood for "low/medium/high", and 'v' was for "very high" (quality). Looks like the man page agrees:

Code:
       rate [-q|-l|-m|-h|-v] [override-options] RATE[k]
              Change the audio sampling rate (i.e. resample the audio) to any given RATE (even non-integer if
              this is supported by the output file format) using a quality level defined as follows:

                                          Quality   Band-   Rej dB   Typical Use
                                                    width
                                    -q     quick     n/a    ≈30 @    playback on
                                                             Fs/4    ancient hardware
                                    -l      low      80%     100     playback on old
                                                                     hardware
                                    -m    medium     95%     100     audio playback
                                    -h     high      95%     125     16-bit mastering
                                                                     (use with dither)
                                    -v   very high   95%     175     24-bit mastering

              where  Band-width is the percentage of the audio frequency band that is preserved and Rej dB is
              the level of noise rejection.  Increasing levels of resampling quality come at the  expense  of
              increasing  amounts  of  time to process the audio.  If no quality option is given, the quality
              level used is `high' (but see `Playing & Recording Audio' above regarding playback).

              The `quick' algorithm uses cubic interpolation; all others use band-limited interpolation.   By
              default,  all  algorithms have a `linear' phase response; for `medium', `high' and `very high',
              the phase response is configurable (see below).

              The rate effect is invoked automatically if SoX's -r option specifies a rate that is  different
              to that of the input file(s).  Alternatively, if this effect is given explicitly, then SoX's -r
              option need not be given.  For example, the following two commands are equivalent:
                 sox input.wav -r 48k output.wav bass -b 24
                 sox input.wav        output.wav bass -b 24 rate 48k
              though the second command is more flexible as it allows rate options to be  given,  and  allows
              the effects to be ordered arbitrarily.

                                                    *        *        

Top
#368360 - 17/02/2017 20:22 Re: Empeg and Rio guys: question about 48khz sample rates on Empeg etc [Re: tfabris]
mlord
carpal tunnel

Registered: 29/08/2000
Posts: 13443
Loc: Canada
I'll take an existing mp3 from the archives here, and use sox to re-encode it at 48KHz, and with luck find out what one of my players thinks about the result..

Top
#368361 - 17/02/2017 20:28 Re: Empeg and Rio guys: question about 48khz sample rates on Empeg etc [Re: mlord]
tfabris
carpal tunnel

Registered: 20/12/1999
Posts: 30505
Loc: Seattle, WA
Ah, I see.

I'm on Windows, so I don't have a "man" command to use.

Instead, I was using the only alternative I had: Their link to the main sox manual from their documentation page on SourceForge.

This appears to be the root of my documentation problem. Your man page is better than what's up there. For instance, that critical table of parameters:
Code:
                                          Quality   Band-   Rej dB   Typical Use
                                                    width
                                    -q     quick     n/a    ≈30 @    playback on
                                                             Fs/4    ancient hardware
                                    -l      low      80%     100     playback on old
                                                                     hardware
                                    -m    medium     95%     100     audio playback
                                    -h     high      95%     125     16-bit mastering
                                                                     (use with dither)
                                    -v   very high   95%     175     24-bit mastering


... that table is completely missing from their documentation page on the SourceForge web site. So it's no wonder that I couldn't figure it out: They literally didn't have the most important part included in their own documentation. And, as you said, your man page there has the correct character for hyphens, unlike their web page. So what you just posted there is super helpful.

Can you find the corresponding section of the man page for the options for controlling MP3 encoding? That also seems to be completely missing from their docs on the SourceForge site.
_________________________
Tony Fabris

Top
#368362 - 17/02/2017 20:41 Re: Empeg and Rio guys: question about 48khz sample rates on Empeg etc [Re: tfabris]
tfabris
carpal tunnel

Registered: 20/12/1999
Posts: 30505
Loc: Seattle, WA
I see that the PDF version of the docs includes the samplerate conversion settings information that was missing from their HTML version of the docs. So the root of the problem is that the HTML version of their doc page has some kind of a malfunction where it is dropping text.

Proof: Search for the following text:
Code:
24-bit mastering

Search in each of these pages:
http://sox.sourceforge.net/sox.html
http://sox.sourceforge.net/sox.pdf

Note that the text only appears in the latter page.

But even with the PDF version, I still can't find anything on how to control the MP3 encoding bitrate or quality level. Can you find it in the man page?
_________________________
Tony Fabris

Top
#368363 - 17/02/2017 20:47 Re: Empeg and Rio guys: question about 48khz sample rates on Empeg etc [Re: mlord]
tfabris
carpal tunnel

Registered: 20/12/1999
Posts: 30505
Loc: Seattle, WA
Originally Posted By: mlord
I'll take an existing mp3 from the archives here, and use sox to re-encode it at 48KHz, and with luck find out what one of my players thinks about the result..


Great idea! Remember that the distortion caused by poor samplerate conversion is in the range that might be considered "subtle" for some people unaccustomed to listening for audio quality. Make sure to use an MP3 which would clearly show the distortion to your ears. For instance, songs with distorted guitars are probably not good for this test. Better if it's something clean, like a quiet acoustic piano piece with a lot of long note decays, or a solo a-capella vocal with some nice held-out vowels. And it should be something you know well, so that the difference would be noticeable to you.
_________________________
Tony Fabris

Top
#368364 - 17/02/2017 21:11 Re: Empeg and Rio guys: question about 48khz sample rates on Empeg etc [Re: tfabris]
mlord
carpal tunnel

Registered: 29/08/2000
Posts: 13443
Loc: Canada
Here is the basic syntax I am using:

sox old.mp3 -C 128 new.mp3 rate -v 48k

Thus far, I've done two tracks, and cannot tell the difference.
Probably not picking the "right" track for it though.

EDIT: Found a very suitable track, with clear soaring vocals against a quiet backdrop to open with, and some nice two voice harmony. Sounds fine at both 44.1 and 48.0 on my Empeg Mk2 player. Making the tracks available to Tony now (PM).

-ml


Edited by mlord (17/02/2017 21:57)

Top
#368365 - 17/02/2017 21:29 Re: Empeg and Rio guys: question about 48khz sample rates on Empeg etc [Re: tfabris]
mlord
carpal tunnel

Registered: 29/08/2000
Posts: 13443
Loc: Canada
Note that the secret "-C 128" flag, given between the two files, is what specifies the resulting .mp3 bitrate. I had to google for that.. not in the man page here.

SoX is an amazing professional tool. With incredibly difficult documentation. smile

Top
#368366 - 17/02/2017 22:02 Re: Empeg and Rio guys: question about 48khz sample rates on Empeg etc [Re: tfabris]
mlord
carpal tunnel

Registered: 29/08/2000
Posts: 13443
Loc: Canada
Oh, and thank-you Tony for getting me to hook up and fiddle with an empeg again. Haven't really done much with them here since The Move 2 years ago. The player I'm fiddling with now is the same one that I got prompted to put a 64GB SSD into a few weeks back. smile

It'll end up as the workshop jukebox after I load it up a bit more fully later on.

Cheers!

Top
#368367 - 17/02/2017 22:35 Re: Empeg and Rio guys: question about 48khz sample rates on Empeg etc [Re: mlord]
tfabris
carpal tunnel

Registered: 20/12/1999
Posts: 30505
Loc: Seattle, WA
I won't be able to check out the files on the empeg until Saturday/Sunday ish, but I'll do that. Thanks! And thanks for the information about the MP3 bit rate command line parameter.

I'm considering putting a larger SSD into mine, too. 64gb would be just about the right size. I'll either find an existing SSD thread or start a new one on that topic.
_________________________
Tony Fabris

Top
#368375 - Yesterday at 18:13 Re: Empeg and Rio guys: question about 48khz sample rates on Empeg etc [Re: tfabris]
tfabris
carpal tunnel

Registered: 20/12/1999
Posts: 30505
Loc: Seattle, WA
Okay: Results of the tests:

Mark's supplied MP3 files from his collection:
- Mark's 44.1k file: Good: No distortion.
- Mark's 48.0k file: Bad: Distortion.

My files:
- Original 48.0k file: Bad: Distortion.
- Convert Original file to 44.1k: Good: No distortion.
- Convert (redo) that 44.1k file back to 48.0k: Bad: Distortion.

(By the way, my command lines for the back and forth conversion I described above were:
sox.exe Lincoln48.mp3 -C 360 Lincoln44.1.mp3 rate -v 44.1k
sox.exe Lincoln44.1.mp3 -C 360 Lincoln48redo.mp3 rate -v 48k
Also, I checked all files in Winamps "file properties" box and made sure it agreed with the sample rates, and made sure that the track names were carefully tagged so that I could tell for sure which one is which during playback on the Empeg.)

So it's definitely a problem that is specific to the sample rate, and it is not a quirk of the encoding. It's also definitely not a bitrate thing, because on my files I was working exclusively at high bit rates (now that I know the Sox command line for it), and Mark's files were at 128k.

Mark says he couldn't hear the difference in the files he supplied, but I could hear it. It's a subtle difference, and I'd only say the difference was "clear" at certain points in the songs where there was content which made it obvious (in this case a solo female voice singing open vowels). But at those points the distortion was clear as day to me, but I could see how it would be subtle to pick out for some folks.

There's also the possibility that there's a difference in firmware or hardware on Mark's empeg from mine, and that the version he's running is better about 48khz playback.
_________________________
Tony Fabris

Top
#368377 - Today at 00:42 Re: Empeg and Rio guys: question about 48khz sample rates on Empeg etc [Re: tfabris]
mlord
carpal tunnel

Registered: 29/08/2000
Posts: 13443
Loc: Canada
Great! I totally trust your (semi-pro musician's) ear on this one. Subtle difference though, but if you hear it then it is there.

Enjoy the tune I sent? smile

Cheers!

Top
#368379 - Today at 00:50 Re: Empeg and Rio guys: question about 48khz sample rates on Empeg etc [Re: tfabris]
mlord
carpal tunnel

Registered: 29/08/2000
Posts: 13443
Loc: Canada
One thought: does your empeg have the "capacitor fix" applied, or is it new enough (RioCar) not to require it?

Top