Korg Forums A forum for Korg product users and musicians around the world. Moderated Independently. Owned by Irish Acts Recording Studio & hosted by KORG USA
The problem on Pa3XLe is not the mp3 codec because if only the voice is recorded without harmony is pristine quality. The problem starts when the CPU/DSP overloads using the harmonizer which leads to issues on the voice recording.
The first video shows the harmonizer factory Korg setup quality without any circuit modification on mic input. Notice the voice "bubbling".... but It was a first learning very exited to have a Pa3XLe in my studio.
I corrected some of the recording issues adding an anti-aliasing ADC filter on mic input. On next video with mic anti-aliasing filtering on the bridge part with harmony activated the problem of CPU DSP overload appears like voice "bubbling", the issue is random. Here I'm using an external effects processor a Sony MP5 for the voice "tape echo" effect.
On next video the voice quality is crystal clear using a better recalculated ADC anti-aliasing filter without external effects but voice "bubbling" on harmony part still remains. The mic is a Prodipe TT1 Pro Lanen.
Next video shows the quality of voice without harmony using a much better LCR anti-aliassing ADC filter. Now the voice is top pristine but when TC-Helicon harmony is "on" the leveling of the main voice changes like "bubbling", anyway works very well being a mp3 recording with the ADC anti-aliasing filter.
Despite my way of singing and wrong english when DM3725 CPU/DSP overloads and the TC-Helicon harmonizer is in use a sort of voice "bubbling" appears. And in some circumstances the sequence freeze when the CPU/DSP overloads to much. On some mp3 recordings the harmony issue is like a "ping-pong delay" of the voice from left to right like panning fast... very strange the behavior on Pa3XLe during recordings. _________________ Please check my Facebook page and blog at
Researching on youtube a video about the inside on Pa4X from left to right it seems to have a true TC-Helicon DSP processor because there are two RAM chips located on the jacks board below the MIDI connectors.
At first conclusion means that the Pa4X use an additional DSP for the voice harmonizer.... ??
Another video on youtube about Pa4X running TC-Helicon effects the audio quality sounds very good because the digital compression of the voice is "true". Running a separate DSP the limitations on Pa3XLe voice processing are avoided.
The audio quality on Pa4X is far superior from Pa3XLe nothing to say.... _________________ Please check my Facebook page and blog at
jeez!!!! enough already!!!
Pete, I think it might be time to get the hammer out
Hmmmm ...but I only have an older low-resolution hammer
This is becoming somewhat amusing now...... is "amusing" the right word?
...but if this is important to some owners, then it's fine I guess.....
P _________________ PA4X-76, Karma, WaveDrum GE, Fantom 8 EX
------------------------------------------------------------------
## Please stay safe ##
...and play lots of music
------------------------------------------------------------------
Low resolution hammer is more solid and more effective because of bigger iron pixels concentration but it seems that topic
owner has already understood the obvious about Pa4X and the rest will exclusively concern Pa3XLe section ... _________________ Music Conductor - Sound Engineer & Developer - Automotive SMPS/RF R&D - Electronics Engineer PaSeries Demos - WavesArt.eu - KorgPa.gr <> Facebook Keyboards : Steinway-D, Kronos X, Pa5X 76, Pa4X 76, Montage M7 , Roland-XV88, Emu3,Emax II,Synclavier II,Yamaha DX Series, ΟΒ-8V
I open this post to know opinions about the mp3 voice recording quality on Pa4X because the mainboard is the same on all models from entry level Pa600 to flagship Pa4X.
Pa3XLe suffers from CPU/DSP overloading using the harmonizer and mp3 recorder this means that on Pa4X CPU/DSP resources are also limited but the TC-Helicon seems to use a second DSP "true engine" on the jacks board the good news.
The Pa3XLe mic compressor is not so good acts like a "signal limiter" is not a true voice compression but on Pa4X the recorded voice quality is much better with true compression and deeper effects thanks to its second DSP chip such feature is very important. _________________ Please check my Facebook page and blog at
I open this post to know opinions about the mp3 voice recording quality on Pa4X because the mainboard is the same on all models from entry level Pa600 to flagship Pa4X.
Pa3XLe suffers from CPU/DSP overloading using the harmonizer and mp3 recorder this means that on Pa4X CPU/DSP resources are also limited but the TC-Helicon seems to use a second DSP "true engine" on the jacks board the good news.
The Pa3XLe mic compressor is not so good acts like a "signal limiter" is not a true voice compression but on Pa4X the recorded voice quality is much better with true compression and deeper effects thanks to its second DSP chip such feature is very important.
The pa4x is the only one that has a seperate TC-Helicon DSP..pa700 and pa1000 don't have it according to the service manuals..
I have a question though for you..You are constantly talking about overloading of the dsp/cpu..Lets leave aside that it would make clicks/pops if that was the case..The question is, does the quality you hear on the speakers change if you record an mp3 versus when you are not recording?
I believe it doesn't change..
To make everything clear you could try recording to pc (from the analog output jacks to your audio interface) and to mp3 with the built in mp3 recording in the same time and then compare the sound on both...
If they are different sounding then the problem is the mp3 codec and not any overloading of the dsp/cpu and if they are the same record to pc without recording mp3 to keyboard (with all the settings the same on keyboard and audio interface) and compare the 2 files recorded on pc..They should be the same too...
The pa4x is the only one that has a seperate TC-Helicon DSP..pa700 and pa1000 don't have it according to the service manuals..
I have a question though for you..You are constantly talking about overloading of the dsp/cpu..Lets leave aside that it would make clicks/pops if that was the case..The question is, does the quality you hear on the speakers change if you record an mp3 versus when you are not recording?
I believe it doesn't change..
To make everything clear you could try recording to pc (from the analog output jacks to your audio interface) and to mp3 with the built in mp3 recording in the same time and then compare the sound on both...
If they are different sounding then the problem is the mp3 codec and not any overloading of the dsp/cpu and if they are the same record to pc without recording mp3 to keyboard (with all the settings the same on keyboard and audio interface) and compare the 2 files recorded on pc..They should be the same too...
Today I tested again the problem is not the mp3 codec is the lack of enough DSP power which leads to recording issues only on the voice recording. Because the DSP is divided having 8 units resources are dynamic allocated depending the tasks to run is not related to sounds clicks or pops on recordings. In other words if a DSP unit overloads will not affect the rest.. Texas Instruments DM3725 is not a spectacular super power chip costs only 12 dollars because is a single core ARM+DSP with its limitations for a professional arranger. For example the Amazon Alexa Echo a virtual voice assistant use the single core DM3725 meaning that the chip is an entry level ARM/DSP.
The Pa4X improves the recording quality thanks to its true TC-Helicon DSP but Pa3XLe, Pa900 and Pa1000 suffers the issues exposed here. _________________ Please check my Facebook page and blog at
To close so everyone can understand this post I recorded right now with the Pa3XLe without mic circuit modification at maximum mp3 quality. Notice the artifacts and issues on voice recording a sort of ping-pong delay from left to right and the artifacts on TC-Helicon harmonizer. This happens when "TC-Helicon is on" like a DM3725 DSP overloading issue with artifacts a total crap the quality.
The backing chord arranging sequence runs flawlessly only the voice is affected meaning that the mp3 codec is not the issue I can hear the voice artifacts on headphones during recording. On some cases during long recordings the arranging sequence start to freeze.
Using a ADC antialiasing filter improves a lot the recording quality but some issues remains. _________________ Please check my Facebook page and blog at
That VP sound is terrible - in fact (seriously) sounds like the Genos VP and reminds me of my Tyros and PSRs910 days.
If my PA4X sounded like that, I'd sell it!!
I use the PA4X VP in every song in my band and I love the quality but it is nothing at all like your video.
Pete _________________ PA4X-76, Karma, WaveDrum GE, Fantom 8 EX
------------------------------------------------------------------
## Please stay safe ##
...and play lots of music
------------------------------------------------------------------
That VP sound is terrible - in fact (seriously) sounds like the Genos VP and reminds me of my Tyros and PSRs910 days.
If my PA4X sounded like that, I'd sell it!!
I use the PA4X VP in every song in my band and I love the quality but it is nothing at all like your video.
It seems that Yamaha Genos use maybe a Texas Instruments DM3725 because vocal processing sounds equal as Pa3XLe. Same DSP coding same libraries same issues nothing changes on brands. Here I recorded a new video about DM3725 internal DSP overload on Pa3XLe. The audio stream started to cut at 0:48 (-0:13) a part is lost during the recording.
Tom Lang gives an introduction about the Pa3XLe vocal harmony. To avoid issues is important not to overload the mic singing loud because of the lag of signal compression on Pa3XLe. Following such rules the quality sound acceptable.
Yamaha Tyros sounds great because use a second vocal processing DSP the former technology without issues.
The introduction of cheap single core ARM+DSP all in one solution for tablets ans smartphones like the DM3725 is not a progress at all. In Yamaha Genos (DSP jittering issues starting at 1:12) and Korg Pa series are the culprit about the exposed issues.
_________________ Please check my Facebook page and blog at
Is not a question of customer support because we are talking about design flaws. Making configurations changes via phone calls does´n helps but researching youtube videos yes here is a demo about the Pa3X Le. It´s funny and shows what can be done with the Pa3XLe the same quality in arranging styles as the Pa4X.
This inspired me to test in another way the TC-Helicon on Pa3XLe it works very well if the arranging sequence is not complex and the voice matches in harmony for example a simple Jazz Bebop style. Recorded using the on board mp3 function.
And sound very good without harmony issues the singer voice match is the key on Pa3X Le... And for only 1500 dollars new running a DM3725 single core tablet ARM+DSP chip the Pa3XLe is a great arranger. _________________ Please check my Facebook page and blog at
Joined: 02 May 2008 Posts: 3990 Location: Los Angeles
Posted: Tue Aug 21, 2018 12:57 am Post subject:
"Is not a question of customer support because we are talking about design flaws...."
If you're so convinced that your keyboard fails to deliver as advertised, why wouldn't you want to get in touch with them and present the evidence? What's the point of doing a technical research without coming to a conclusion? What's the point of all these 4 pages then? _________________ Sam
It seems that Yamaha Genos use maybe a Texas Instruments DM3725 because vocal processing sounds equal as Pa3XLe. Same DSP coding same libraries same issues nothing changes on brands. Here I recorded a new video about DM3725 internal DSP overload on Pa3XLe. The audio stream started to cut at 0:48 (-0:13) a part is lost during the recording.
The introduction of cheap single core ARM+DSP all in one solution for tablets ans smartphones like the DM3725 is not a progress at all. In Yamaha Genos (DSP jittering issues starting at 1:12) and Korg Pa series are the culprit about the exposed issues.
Yamaha is using yamaha propriety chips amd processors..
They build and manufactor them theirselves..
Which makes your conclusions and assumtions utter nonsense... _________________ Korg Kronos 2/88 , Genos, Mainstage3 +VSTsu, ipad pro, GSi Gemini, Roland Integra 7, Jupiter Xm, Yamaha motif XS rack, Ketron SD90.
www.keyszone.boards.net
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum