MP3 songs record on PA4x not playable on Tyros 5

AntonySharmman · Post by **AntonySharmman** » Tue Dec 01, 2015 8:29 pm

One thing is totally inaccurate in that older post !
A variable bit rate Mp3 encoding is determined by the highest bit that this algorithm works , eg when a file is signed as 128 Kbps
VBR then maximum value is 128 Kbps , only Average bitrate (ABR) encoding has as an average bit rate 128 kbps with min. 64 &
max. 256 kbps .. I was working for years with Fraunhofer encoding (world's best ever encoder) !

But you all have missed the main issue here that I've already posted ... The above uploaded by byme88 file isn't either VBR
nor ABR Mp3 format ... it's a 32 Kbps CBR (Constant Bit Rate ) garbage MP3 with corrupted metadata that any professional
Sound Editor can confirm and has nothing to do with T4 or other players , so you'd better explore what is wrong in Mp3 algorithm
programming in Pa4X , if this uploaded file is indeed valid ...

Paolo@Korg · Post by **Paolo@Korg** » Tue Dec 01, 2015 9:56 pm

Antony,

These are the info I get from a few audio applications:

QuickTime Player 7: 234.38 kbit/s 16 bit
Amadeus: 192 kbps 16 bit
Adobe Audition: 32 kbps 32-bit float
Audacity: 32-bit float
iTunes: 32 kbps
VLC: 32 kbps

Each one has its method of reading the bit rate of a variable-rate file, that is, something that by definition has no fixed rate. The apps showing 32 kbps are probably reading the beginning of the file; 32 kbps is a quite obvious compression for silence.

All the above apps read and play the file with no problems. While they don't give me detailed information on the encoding of the source file, Logic, GarageBand, SonicVisualizer, Praat, all can read this file.

I would say that the problem is in the only device that cannot read it.

Paolo

byme88 · Post by **byme88** » Tue Dec 01, 2015 10:27 pm

Paolo
Thanks so much for your input!

Other than these minor issues which I can live with, I am overall very please with the PA4x!!!

I will try shortened the file name as well a starting the recording right after the song start to see if it help using the higher bit rates and let you know my findings!

Thanks again!

byme88 · Post by **byme88** » Tue Dec 01, 2015 11:19 pm

AntonySharmman wrote:One thing is totally inaccurate in that older post !
A variable bit rate Mp3 encoding is determined by the highest bit that this algorithm works , eg when a file is signed as 128 Kbps
VBR then maximum value is 128 Kbps , only Average bitrate (ABR) encoding has as an average bit rate 128 kbps with min. 64 &
max. 256 kbps .. I was working for years with Fraunhofer encoding (world's best ever encoder) !

But you all have missed the main issue here that I've already posted ... The above uploaded by byme88 file isn't either VBR
nor ABR Mp3 format ... it's a 32 Kbps CBR (Constant Bit Rate ) garbage MP3 with corrupted metadata that any professional
Sound Editor can confirm and has nothing to do with T4 or other players , so you'd better explore what is wrong in Mp3 algorithm
programming in Pa4X , if this uploaded file is indeed valid ...

The uploaded file is indeed 100% valid recorded from the PA4x.... Thanks!

AntonySharmman · Post by **AntonySharmman** » Tue Dec 01, 2015 11:32 pm

Well Paolo , I have to disagree in one significant point , defending only the pro utilities I work with , Adobe , Waves Lab & Audacity
telling you that all are reading by definition the attached in every mp3 file first lines code that is called metadata where all file
information are announced plus some extra sound information useful for decoding this file back to a lossless format.
That you've missed Paolo is that all Editors define this file as CBR (Constant Bit Rate) ...

In case that this metadata info is missing or unreadable then those utilities will decide to provide the info either based on average bit
rate of whole file nor by the result of the very first byte of bit rate decoding that is used there and seems very reasonable but this
is not their fault !
On the other hand , an older Adobe Audition v5.5 that is mainly working under Fraunhofer source code encoding gives the actual error here ...

When tested this file with spectrum analyser , I found frequency information up to 16 KHz , which means that this
file definitely isn't 32 kbps CBR but more than 128 Kbps that also leads to bad metadata conclusion.

My opinion is to point KorgPa programmers at this Pa4X Mp3 recording (Pa3X Mp3 had never any issues) due to
test created metadata for any case ...

Greetings

byme88 · Post by **byme88** » Wed Dec 02, 2015 6:28 am

Paolo,

I renamed the file and still could not play on the Tyros 5.
But, if I start recording on PA4x with the song start first, and saved it with a shorter name then I can play it on the T5 and T4!!!!!

So the solution is not to record it with silence to start!

It is the bug on the PA4x and it need to be addressed!

Here is the link to the file recorded right after the song started:

https://www.dropbox.com/s/3kdm4ikgab8cgrb/TACE.MP3?dl=0

Thanks,

Reuben · Post by **Reuben** » Wed Dec 02, 2015 6:45 am

I recently tried to play/import an mp3 recorded on my Korg Pa3x in the iPad GarageBand app and it wouldn't play.
The solution to fix it was to rename the extension from capital MP3 to lower case mp3 and it worked immediately.
I wonder if something like this could be the problem??
John

AntonySharmman · Post by **AntonySharmman** » Wed Dec 02, 2015 11:16 am

This is the result of new TACE.MP3 file even when renamed back to "khi em thoanh qua doi toi.MP3"

Therefore the name is independent from format recognition , it's presented as 233 Kbps VBR this time but metadata issue warning still exists !
This IMHO requires further investigation !!!

alfredokiwi · Post by **alfredokiwi** » Fri Sep 21, 2018 8:23 pm

Are such issues corrected using a better mp3 encoder on latest Pa4X OS update ?? Because on Pa3XLe audio quality is poor it seems to be at low bit rate going from 32kbs to 64kbs average. Also the mp3 metadata seems to be wrong because reports an incorrect duration playing back the files in a PC.
The encoder type reported by the application Audio Identifier is FhG the Fraunhofer mp3 encoder being recorded at "bad" poor quality, see pic link below

https://www.facebook.com/photo.php?fbid ... =3&theater

I switch to a fast USB 3.1 memory stick a Kingston DT50 16Gb which seems to resolve the "audio cuts" but mp3 quality issues remains during recordings.

On wikipedia seems to be FhG recording at high quality despite the reported low bit rate by Audio Identifier https://wiki.hydrogenaud.io/index.php?t ... er_FDK_AAC

The bitrate limit for each variable bitrate mode. [12] HE and HEv2 will often end up with actual bitrates far below these limits.

AACENC_BITRATEMODE

Mono Stereo
1 VBR 32 kbps 20 kbps LC, HE, HEv2
2 VBR 40 kbps 32 kbps LC, HE, HEv2
3 VBR 56 kbps 48 kbps LC, HE, HEv2
4 VBR 72 kbps 64 kbps LC
5 VBR 112 kbps 96 kbps LC
------------------------------------------------------------------------

AACENC_BITRATEMODE

1 13050 Hz
2 13050 Hz
3 14260 Hz
4 15500 Hz
5 Full range, no filter

Being bitrate mode 5 at 96kbps stereo the maximal quality.

alfredokiwi · Post by **alfredokiwi** » Sat Sep 22, 2018 12:05 am

I have verified again on Pa3XLe the mp3 encoder seems to record at:

160 Kbs 48khz at medium quality
192 Kbs 48Khz at high quality
256 Kbs 48Khz at highest quality

Reported by Mp3 Doctor. Recording quality from my point of view is affected by the signal level variation on internal digital DSP bus. Such issues are happening more on the mic voice it seems that during mp3 recording the normalization process sometimes fails leading to fluctuations on the recording.

Because the Pa4X has a separate true TC-Helicon engine the digital voice compression is accurate avoiding the issues that happens on Pa3XLe TC-Helicon voice recording.

https://www.facebook.com/photo.php?fbid ... =3&theater

From my point of view the Fraunhofer mp3 encoder "FhG" on Pa series is not the culprit.

alfredokiwi · Post by **alfredokiwi** » Sat Sep 22, 2018 12:32 am

Korg Pa3XLe mp3 recording issue root cause resolved !!!I The problem on Pa series mp3 recordings appears when you start the recording without sound (internal DSP bus without signal). This enforces the mp3 plugin to set the lowest bitrate. To avoid issues is important to run the arranger to feed signal to DSP bus before recording, stop and start again without stopping the recording. Such procedure will set the mp3 bitrate to its correct values on mp3 encoder.
Below is the picture about the issue, the last 3 files are without sound nothing started which sets the Fraunhofer mp3 encoder "FhG" at 32Kbs lowest rate.

https://www.facebook.com/photo.php?fbid ... =3&theater

To resolve the issue on the next OS update the FhG encoder should be configured at CBR 128, 192, 256 depending on the recording quality selected by menu (medium. high, highest) avoiding the use of VBR encoding.

alfredokiwi · Post by **alfredokiwi** » Sat Sep 22, 2018 2:17 am

Again the Korg arrangers Pa series mp3 recording issue is resolved feeding the DSP bus with a high frequency signal before recording. A white noise signal generator using for example your voice singing a shyyyyyyyyyyyy (similar to radio FM noise) before starting the mp3 recording works pretty well because has frequency components in all the spectrum telling to the mp3 encoder to record at proper quality. I tested it right now such workaround fix the problem the green files are the correct mp3 recordings following such procedure.

<iframe src="https://www.facebook.com/plugins/video. ... &width=560" width="560" height="420" scrolling="no" frameborder="0" allowTransparency="true" allowFullScreen="true"></iframe>

Also the length of the recording is properly set when the file is open in a PC. Using the singing "shyyyyyyyyy" white noise method to feed the DSP with signal is a crazy idea that works. The loud voice is properly recorded the harmony with some issues but pass. What remains to do is a proper setup of the voice mic input level, the EQ and the limiter configuration to avoid signal overload on the internal DSP bus during recordings.

https://www.facebook.com/photo.php?fbid ... =3&theater

karmathanever · Post by **karmathanever** » Sat Sep 22, 2018 8:09 am

Tyros-5 tech wrote:The T5 is fairly easy going concerning the mp3 bit rate of files (64-320 kbps), but it will only play those sampled at 44.1 or 48.0 kHz.