Equalizing a Digital Piano – Need Advice!

[Tallulah]

I have a digital piano, but at the moment, I can only use it with headphones because my room has poor acoustics—particularly an issue at 130 Hz that coincides with the C3 note. Implementing acoustic treatment is not a viable option due to financial and space constraints.

For this reason, I’ve decided to generate an equalization file with REW for my digital piano’s internal speakers. While this might seem like a risky idea, it is very likely to solve my problem. To apply the equalization, I would connect the piano to my PC via MIDI, use a DAW with a VST such as Kontakt or Pianoteq, apply the equalization with ReaVerb (if there’s a better option, I’d appreciate any suggestions), and then send the already equalized sound back to the piano’s speakers via ASIO to minimize latency.

This is my first time using REW. I’ve purchased a UMIK-1, although I’m still waiting for it to arrive.

I’ve done some research, but I’m quite new to this topic and would love to hear the opinions of those with more experience. Right now, two questions are constantly on my mind:

1. My digital piano is a console-style model, and its speakers are located at the bottom, facing the floor. I know the microphone should be placed roughly in the listening area, but what about its orientation? Should it point forward with the 0° calibration, or would it be more appropriate to use the 90° calibration and point it toward the ceiling? I have many doubts due to the speaker positioning.

2. Should I measure the left and right channels separately and generate a different equalization file for each, or would it be better to measure both channels together and use a single equalization file for both?

I’ve searched for information online, but since my case is somewhat specific, I’d rather ask in the forum. Any guidance or suggestions would be greatly appreciated!

 

[GXAlan]

It’ll be hard to do this digitally without introducing latency. Your best bet is to use a sample that mirrors your intended tilt. (Brighter vs darker).

In many cases, even the latency of a ADC/DAC with no processing interrupts your playing and in some cases when you are playing fast, the latency of something like pianoteq is still a problem. For casual, jazzy tunes it’s probably not as bad. I don’t know if pure modeling pianos allow you EQ without latency. As popular as VST’s are, there are those who don’t like it even as stock. I would ask on pianoworld if there are faster VST’s to minimize latency.

Keep in mind that pianos aren’t flat to a standard even at the highest performance levels like a contracted Steinway Artist.

1) I would use the 90° calibration. The 0 degree calibration is giving your on axis response but for a piano like that, you are probably hearing a lot more reverberation.

2. I probably would recommend independent EQ.

Just as a test, I would also try to EQ at the audience position and seeing how that affects how you experience the sound at the seated position.

When you are done, it might be fun to compare the speaker performance to this Korg.

KORG Poetry (Digital Piano/Bluetooth Speaker) Measurements

KORG Poetry Digital Piano/Soundbar The KORG Poetry is a digital piano which has been designed as "The Piano For Lovers of Chopin." It uniquely captures an 1843 Pleyel vintage piano that was Chopin's choice of piano and also provides a modern Fazioli grand piano sample. What's unique is that...

www.audiosciencereview.com

 

[TurtlePaul]

In practice the 90 degree or zero degree wont matter at all for fixing 130 hz.

 

[Tallulah]

It’ll be hard to do this digitally without introducing latency. Your best bet is to use a sample that mirrors your intended tilt. (Brighter vs darker).

In many cases, even the latency of a ADC/DAC with no processing interrupts your playing and in some cases when you are playing fast, the latency of something like pianoteq is still a problem. For casual, jazzy tunes it’s probably not as bad. I don’t know if pure modeling pianos allow you EQ without latency. As popular as VST’s are, there are those who don’t like it even as stock. I would ask on pianoworld if there are faster VST’s to minimize latency.

Keep in mind that pianos aren’t flat to a standard even at the highest performance levels like a contracted Steinway Artist.

1) I would use the 90° calibration. The 0 degree calibration is giving your on axis response but for a piano like that, you are probably hearing a lot more reverberation.

2. I probably would recommend independent EQ.

Just as a test, I would also try to EQ at the audience position and seeing how that affects how you experience the sound at the seated position.

When you are done, it might be fun to compare the speaker performance to this Korg.

KORG Poetry (Digital Piano/Bluetooth Speaker) Measurements

KORG Poetry Digital Piano/Soundbar The KORG Poetry is a digital piano which has been designed as "The Piano For Lovers of Chopin." It uniquely captures an 1843 Pleyel vintage piano that was Chopin's choice of piano and also provides a modern Fazioli grand piano sample. What's unique is that...

www.audiosciencereview.com

Thank you for your quick and insightful response. I have been conducting some tests, and while it is true that without ASIO, the delay prevents me from playing accurately, when using ASIO with a small buffer, I haven't noticed any significant discomfort due to latency.

Regarding completely flattening the response, I understand your point. However, I am aware that my room has a couple of exaggerated peaks at very specific frequencies. I don’t have a measurement microphone yet, but I have conducted some experiments with a "standard" microphone I have at home and detected a peak of over 16 dB at 130 Hz. My main goal is to correct these issues caused by the room’s acoustics.

Would you also recommend taking multiple measurements per channel and averaging them? Regarding your comment about sitting in the audience position, that’s not possible since the piano is placed against a wall (which likely affects the acoustics, but I have no other option in this space).

I will share all the measurements as soon as I have the proper microphone!

 

[GXAlan]

How do you input into your piano? Bluetooth, line in? When you use the VST, it goes back to the line in of your piano speakers? (MIDI out to line in)? Could you buy small studio monitors and mount them to the wall and bypass your piano’s speakers?

The other wildcard would be something like an 80’s era equalizers. The Technics SH-8065 comes to mind, but I am not sure I understand how your digital piano connects to your piano speakers.

The Korg has a wire you could splice out, and there is a way to hardwire a line-in. Honestly, I went from a premium stage piano to the Korg because it was good enough. The weighted element of the keyboard is artificial, but it works overall.

 

[Tallulah]

How do you input into your piano? Bluetooth, line in? When you use the VST, it goes back to the line in of your piano speakers? (MIDI out to line in)? Could you buy small studio monitors and mount them to the wall and bypass your piano’s speakers?

The other wildcard would be something like an 80’s era equalizers. The Technics SH-8065 comes to mind, but I am not sure I understand how your digital piano connects to your piano speakers.

The Korg has a wire you could splice out, and there is a way to hardwire a line-in. Honestly, I went from a premium stage piano to the Korg because it was good enough. The weighted element of the keyboard is artificial, but it works overall.

My piano, unfortunately, does not have Line In or Line Out, only a USB cable. It is a Yamaha CLP-825; as the most affordable model in the range, it has limited features. However, fortunately, the USB connection not only transmits MIDI but also functions as an audio interface, allowing the piano's sound to be sent to the PC and vice versa. (On my PC, the piano appears as both an input and an output device.)

Additionally, this USB audio transmission from the Yamaha piano has its own ASIO driver (Yamaha Steinberg USB ASIO), which enables low-latency operation. Currently, my setup works as follows: the piano sends MIDI to the PC, the PC—using a DAW and a VST—synthesizes the sound, applies equalization, and sends the audio back to the piano via USB. By using ASIO, latency is not an issue (at least for me).

I have considered adding studio monitors, but for now, I prefer not to introduce more elements or spend extra money. First, I will run tests using the piano’s built-in speakers.

 

[kemmler3D]

For this project it's important to distinguish between room EQ and speaker EQ.

Room EQ is good for correcting problems at low frequency like you've mentioned. You measure at the listening position.

Speaker EQ is done by taking pseudo-anechoic measurements, usually at multiple angles, to correct problems with the speaker itself independently of the room. This is much more involved and results are variable depending on the limits of the system.

Processing via VST seems fine if you don't mind the latency. I would only EQ aggressively below 300hz or so, and be gentle above that to avoid unexpected results.

Lastly, getting a secondhand pair of JBL 305/8 p will barely cost more than a UMIK and will blow the built-in speakers out of the water no matter how much you EQ the piano. But it won't hurt to try, I guess.

 

[Roland68]

My piano, unfortunately, does not have Line In or Line Out, only a USB cable. It is a Yamaha CLP-825; as the most affordable model in the range, it has limited features. However, fortunately, the USB connection not only transmits MIDI but also functions as an audio interface, allowing the piano's sound to be sent to the PC and vice versa. (On my PC, the piano appears as both an input and an output device.)

Additionally, this USB audio transmission from the Yamaha piano has its own ASIO driver (Yamaha Steinberg USB ASIO), which enables low-latency operation. Currently, my setup works as follows: the piano sends MIDI to the PC, the PC—using a DAW and a VST—synthesizes the sound, applies equalization, and sends the audio back to the piano via USB. By using ASIO, latency is not an issue (at least for me).

I have considered adding studio monitors, but for now, I prefer not to introduce more elements or spend extra money. First, I will run tests using the piano’s built-in speakers.

I don't understand much about it, but if the Yamaha piano has USB output and input, wouldn't an FIR convolver on your PC be ideal?
I've read in several places that there are ways to reduce the latency.

 

[Multicore]

If conversion latency is a concern, why not try something like this, Behringer PEQ2200 or one of its predecessors.

Behringer | Product | PEQ2000

www.behringer.com

https://reverb.com/item/32742523-behringer-peq2200-ultra-q-pro-parametric-equalizer

 

[Tallulah]

You measure at the listening position.

That's right, but my main question is about the microphone's orientation, since the speakers are positioned lower than the mic and are directed towards the floor.

Lastly, getting a secondhand pair of JBL 305/8 p will barely cost more than a UMIK and will blow the built-in speakers out of the water no matter how much you EQ the piano. But it won't hurt to try, I guess.

The microphone cost €99. I just looked up a pair of second-hand JBL 305P speakers, and they don’t sell for less than €200 in my country. I’m sure they would sound much better, but for now, they’re not worth the price. First, I want to see if I can properly address my room’s acoustic issues through equalization—after all, I’d encounter the same problems with any speakers.

Wouldn't an FIR convolver on your PC be ideal?

I believe ReaVerb is already functioning as an FIR convolver. Latency seems fine when using ASIO drivers in the DAW.

If conversion latency is a concern, why not try something like this, Behringer PEQ2200 or one of its predecessors.

Thank you for the advice, but I don't think a 5-band equalizer is precise enough to correct issues like standing waves. Additionally, my piano doesn't have a line-in or line-out connection.

 

[staticV3]

1. My digital piano is a console-style model, and its speakers are located at the bottom, facing the floor. I know the microphone should be placed roughly in the listening area, but what about its orientation? Should it point forward with the 0° calibration, or would it be more appropriate to use the 90° calibration and point it toward the ceiling? I have many doubts due to the speaker positioning.

Use the Moving Microphone Method when measuring the in-room frequency response of loudspeakers.

You can google the term and find both a whitepaper and a number of guides on Youtube and ASR.

The paper recommend a microphone calibrated for random incidence (pg. 11). The UMIK's 90° calibration comes closest, so I'd use that.

During the MMM process, the microphone can be pointed randomly, just avoid pointing it directly at the speaker.

Should I measure the left and right channels separately and generate a different equalization file for each, or would it be better to measure both channels together and use a single equalization file for both?

I'd correct the combined L+R response below Schröder and the individual response above.

If you measure L and R separately and they end up almost identically, then it's totally fine to correct just the combined response.

Also, make sure that in REW's Soundcard tab you set the driver type to Java and select the "EXCL: UMIK-1" device and also the EXCL version of your piano.

 

[Philbo King]

I have a digital piano, but at the moment, I can only use it with headphones because my room has poor acoustics—particularly an issue at 130 Hz that coincides with the C3 note. Implementing acoustic treatment is not a viable option due to financial and space constraints.

For this reason, I’ve decided to generate an equalization file with REW for my digital piano’s internal speakers. While this might seem like a risky idea, it is very likely to solve my problem. To apply the equalization, I would connect the piano to my PC via MIDI, use a DAW with a VST such as Kontakt or Pianoteq, apply the equalization with ReaVerb (if there’s a better option, I’d appreciate any suggestions), and then send the already equalized sound back to the piano’s speakers via ASIO to minimize latency.

This is my first time using REW. I’ve purchased a UMIK-1, although I’m still waiting for it to arrive.

I’ve done some research, but I’m quite new to this topic and would love to hear the opinions of those with more experience. Right now, two questions are constantly on my mind:

1. My digital piano is a console-style model, and its speakers are located at the bottom, facing the floor. I know the microphone should be placed roughly in the listening area, but what about its orientation? Should it point forward with the 0° calibration, or would it be more appropriate to use the 90° calibration and point it toward the ceiling? I have many doubts due to the speaker positioning.

2. Should I measure the left and right channels separately and generate a different equalization file for each, or would it be better to measure both channels together and use a single equalization file for both?

I’ve searched for information online, but since my case is somewhat specific, I’d rather ask in the forum. Any guidance or suggestions would be greatly appreciated!

Since you're using Reaper, try ReaEQ instead. It'll do the job with much less latency. It's a PEQ with unlimited bands.

If decor allows, an overstuffed couch or two is a good way to soak up those nasty bass peaks and nodes.

 

[Tallulah]

Use the Moving Microphone Method when measuring the in-room frequency response of loudspeakers.

You can google the term and find both a whitepaper and a number of guides on Youtube and ASR.

The paper recommend a microphone calibrated for random incidence (pg. 11). The UMIK's 90° calibration comes closest, so I'd use that.

During the MMM process, the microphone can be pointed randomly, just avoid pointing it directly at the speaker.


I'd correct the combined L+R response below Schröder and the individual response above.

If you measure L and R separately and they end up almost identically, then it's totally fine to correct just the combined response.

Also, make sure that in REW's Soundcard tab you set the driver type to Java and select the "EXCL: UMIK-1" device and also the EXCL version of your piano.

Thank you very much for the advice. It makes sense to use the Moving Microphone Method that you mentioned, so I will do that. The most challenging part will be avoiding transmitting too many vibrations to the microphone, but I’ll try to do it carefully.

I’ll first try measuring L and R separately and, from there, decide whether it's worth applying any special correction. Again, thank you for the recommendations.

Since you're using Reaper, try ReaEQ instead. It'll do the job with much less latency. It's a PEQ with unlimited bands.

If decor allows, an overstuffed couch or two is a good way to soak up those nasty bass peaks and nodes.

I will keep your advice on using ReaEQ in mind, although I have a question. In the REW PEQ file, I see that the Q value is used, whereas in ReaEQ, octaves are used instead. It seems there is an online converter (https://sengpielaudio.com/calculator-bandwidth.htm). Do you know if converting the values this way would be sufficient and if there is a way to export files that are directly compatible?

 

[Tallulah]

Today, I finally received the UMIK-1 and have been running some tests. In short, the experiment was a success. The piano went from being nearly impossible to play to sounding moderately good, considering the inherent clarity and distortion limitations of having only two speakers at the bottom, pointing toward the floor.

To give you an idea, before the correction, there was a peak of about 16 dB right at the C3 note. Every time I played it, it completely drowned out the rest of the notes.

Procedure​

• I used the moving microphone method, following RandomEar’s thread, to take the measurements.
• I conducted two independent measurements: one for the left channel and another for the right. The difference between the two channels was significant, likely because the piano is not centered in the room and one side is closer to a corner.
• Based on these measurements, I generated a stereo Impulse Response filter (.wav), applying a different correction for each channel.
• To verify the effectiveness of the correction, I loaded the filter and recorded three L+R sweeps with the equalization applied, averaging the results. I wanted to ensure that, despite the channel-specific corrections, the combined sound remained balanced. The results were quite good.

I have to admit, I was surprised by the linearity of the speakers on this Yamaha Clavinova CLP-825. Even though they are positioned at the bottom and point toward the floor, their response is surprisingly consistent. The only real issues are the 16 dB peak at 130 Hz, caused by the room’s acoustics, and the lack of sub-bass (understandable, given that the speakers are only 4.7 inches).

Finally, I loaded the stereo Impulse Response into Reaper’s Monitoring FX using ReaVerb and played the piano with Kontakt/Pianoteq through the ASIO driver. The latency is imperceptible, so using ReaVerb for real-time equalization isn’t an issue.

All in all, I’m very happy with the results!