Hi everybody, I often record young kids playing violin (solo, with piano, or in groups) using a Zoom H4n Pro with the internal mics (small diaphragm condenser used in XY configuration). I'm looking for ways to improve the sound in post, as the small violins are even more scratchy / trebly than full sized violins, and the internal mics of the Zoom emphasize high frequencies.

I use Audacity (I don't have the time to try other DAWs right now). I know nothing about postprocessing so some input would be much appreciated. Here are two short fragments with violin and piano:

https://drive.google.com/file/d/1wsF9rM ... sp=sharing

https://drive.google.com/file/d/1Ol77jV ... sp=sharing

The first is the unprocessed recording with the Zoom was mounted horizontally, about 1.5m from the violinist, slightly higher than the violin, pointing at an angle between the violin and the piano, but closer to the violin. They were positioned in middle of a rather empty rectangular room of about 8m x 5m x 3m, with a lot of reflections.

The second is the same fragment processed with Audacity (my very first attempt at sound processing, besides cut and pasting sections and slowing down tracks): I applied EQ, reversing by eye the frequency curve of the internal mics, which I got from here: https://www.kenrockwell.com/audio/zoom/h4n.htm (this is for the H4n but I assumed the response of the H4n pro is similar), and then I applied 3.8dB amplification (what Audacity gave me by default).

What do you think? Is this a sensible way to go about EQ? (I know I'm not taking into account the room). Should I also apply compression? If so, with what parameters? Some panning? What else should I do?

Of course I should experiment with mic placement and orientation, and room treatment. In this sample it's not my room and I didn't have time to move around the mic.

To be honest, the unprocessed version sounds better to me. The processed version loses too much clarity to my ears. If you really want to try to tame the harshness I'd try some gentle frequency sensitive compression on the frequencies from about 2kHz upwards. Set the threshold so that the quieter notes are left untouched and the ratio to gently tame the harshness rather than eliminate it altogether.

From taking a quick search it appears that Audacity doesn't come with any frequency sensitive compression so it looks you'll need to download a suitable plug-in. I'm not sure what plug-in formats Audacity supports but if it supports VST then I'd suggest downloading Reaplugs from

https://www.reaper.fm/reaplugs/

and using either ReaXComp or ReaFIR in compressor mode.

Definitely what James say - the second file is much worse than the first. The first is pretty usable. Can make a little test later this evening.

If you'll want a more comprehensive DAW for free there is Cakewalk by bandlab which will allow more than Audacity

I do like the eq direction of making the violin less scratchy.. maybe a bit less of it, or you can also try Melda Production MCharacter, so you can enhance "warm" the fundamental frequency and lower the higher ones, and shape the Violin Sound..

Check out the cello example at 3:02
youtube.com/watch?v=uoYp1gpekkE&t=03m02s

Personally I wouldn't use compression in these recordings style to my ears..

Correcting for a mic's or recording's unflat response is one thing and I've done it myself but normally where the response is much worse than here.

Unfortunately music performance groups often have to put up with overly reverberant, harsh sounding venues. On your recording the "short reflections" in the room probably exacerbate the student's playing. If I was recording it I'd try to make the room sound less prominent.

The violin seems to lack the lower tone warmth we hear in a better instrument. We could try and EQ up those tones but that's only possible if they are to some extent there in the first place. Also boosting that band will only exaggerate the same band in the piano. What we perceive as "squeakiness" is probably more the absence of the lower tones to balance the sound out.

Your EQ in sample 2 seem to have too much reduced the "presence" frequencies, from say 2kHz to 5 kHz making the violin sound distant and hollow, while leaving the squeaky higher frequencies above 5 kHz basically unaltered. I would leave the "presence" band basically unaltered, for that carries a lot of the colour of the performance, but play around with a gentle low pass filter starting around 5 kHz or even higher.

Even with that, someone with a good ear will probably still pick that the violin doesnt have a sweet, rounded sound.

All the best with it.

Thanks all for listening to the samples and for the advice. Listening again, I agree that the processed sample lacks clarity.

Tim, indeed smaller violins lack the low frequencies of larger violins. What I'm trying to do is to correct for the coloration of the Zoom, as the sound is consistently thinner than what I ear live standing near the microphone. Actually in this sample this was attenuated by the reverberation. gave it now another try reducing less aggressively the "presence" frequencies:
https://drive.google.com/file/d/17zA2e7 ... sp=sharing

These are the EQ I used in samples 2 and 3:
https://drive.google.com/file/d/1tIENm4 ... sp=sharing

https://drive.google.com/file/d/1S5HgY7 ... sp=sharing

James, Audacity supports VST, I'll look into the plugins you mention.

Tom, I'm not sure I need a more powerful DAW, I hardly know how to take advantage of Audacity's capabilities. I watched the whole video on MCharacter, it sounds like an amazing tool, I may try the trial version for fun.

notuno wrote:Hi everybody, I often record young kids playing violin (solo, with piano, or in groups) using a Zoom H4n Pro with the internal mics (small diaphragm condenser used in XY configuration). I'm looking for ways to improve the sound in post, as the small violins are even more scratchy / trebly than full sized violins, and the internal mics of the Zoom emphasize high frequencies.

To be honest, in a situation like that, I think the wisest course of action is simply to accept that it is what it is... Your recording is a reasonably accurate rendition of the performance and the relative perspectives of violin, piano and room are pretty good.

Yes, you could tame the extreme HF slightly as James has suggested, but I'd say even that is optional and not essential -- it really won't make a substantial overall improvement. Anything more will swallow up hours of post-production time and effort, with no real benefits... and in reality are more likely to make things worse rather than better.

In fact, I'd say there's nothing you can sensibly do to make this performance match the sound you probably have in your head of a maestro playing a Strad because that tonality just isn't there in the source. It's a scratchy trebly violin. That's what it sounds like because that's all its size and construction, and the performance, are capable of producing.

And to my ears, It's not the sonics that draw my attention anyway, it's the tuning and intonation -- which is to be expected and probably accepted from a young (and clearly very promising) student.

notuno wrote:Thanks all for listening to the samples and for the advice. Listening again, I agree that the processed sample lacks clarity.

You can improve the sonics a little bit imho but the room is printed in the recording and there's little to do about that. I did send you a slight worked version yesterday, you should the link in your inbox. The main difficulty was listening to the pitching :D :D but they're students so it's fine :D

CS70 wrote:

notuno wrote:Thanks all for listening to the samples and for the advice. Listening again, I agree that the processed sample lacks clarity.

You can improve the sonics a little bit imho but the room is printed in the recording and there's little to do about that. I did send you a slight worked version yesterday, you should the link in your inbox. The main difficulty was listening to the pitching :D :D but they're students so it's fine :D

Thanks, I missed the message yesterday but I just listened to it now. It's an improvement, did you just cut some high frequencies? This particular fragment is indeed quite out of tune, I should have chosen better...

Hugh Robjohns wrote:

notuno wrote:Hi everybody, I often record young kids playing violin (solo, with piano, or in groups) using a Zoom H4n Pro with the internal mics (small diaphragm condenser used in XY configuration). I'm looking for ways to improve the sound in post, as the small violins are even more scratchy / trebly than full sized violins, and the internal mics of the Zoom emphasize high frequencies.

To be honest, in a situation like that, I think the wisest course of action is simply to accept that it is what it is... Your recording is a reasonably accurate rendition of the performance and the relative perspectives of violin, piano and room are pretty good.

Yes, you could tame the extreme HF slightly as James has suggested, but I'd say even that is optional and not essential -- it really won't make a substantial overall improvement. Anything more will swallow up hours of post-production time and effort, with no real benefits... and in reality are more likely to make things worse rather than better.

In fact, I'd say there's nothing you can sensibly do to make this performance match the sound you probably have in your head of a maestro playing a Strad because that tonality just isn't there in the source. It's a scratchy trebly violin. That's what it sounds like because that's all its size and construction, and the performance, are capable of producing.

And to my ears, It's not the sonics that draw my attention anyway, it's the tuning and intonation -- which is to be expected and probably accepted from a young (and clearly very promising) student.

Thanks. These recordings are mostly for sharing with the kids relatives. Many of them notice the squeaky sound and not so much the intonation problems (evident in this sample from a seven year old).

Until now I was just syncing video with the audio straight from the Zoom. The sound being what it is, I was hoping to find an EQ setting (and maybe compression) to tame the extra harshness due to the microphone, that I could quickly apply to all recordings. Only for the very best performances I would consider investing more time.

As it is not for international release you might consider recording in such a way as to keep piano and fiddle "hard panned" left and right to facilitate separate processing.

I would suggest running subsequent fiddle recordings through an appropriate impulse response in a convolution reverb unit. It's a lot easier than you might think but will obviously affect both instruments unless they can be separated at source.

They can always be pulled back towards the centre later.

I had a play with the file through an impulse response created for an acoustic guitar which was really effective at taming the squeak, unfortunately it didn't improve the intonation. It is a good way to "put in what isn't there" but can be a bit of a rabbit hole once you start experimenting.

None of what I describe is complicated or expensive and this is a good place to find advice on such things.

notuno wrote:

CS70 wrote:

notuno wrote:Thanks all for listening to the samples and for the advice. Listening again, I agree that the processed sample lacks clarity.

You can improve the sonics a little bit imho but the room is printed in the recording and there's little to do about that. I did send you a slight worked version yesterday, you should the link in your inbox. The main difficulty was listening to the pitching :D :D but they're students so it's fine :D

Thanks, I missed the message yesterday but I just listened to it now. It's an improvement, did you just cut some high frequencies? This particular fragment is indeed quite out of tune, I should have chosen better...

Two EQs on that one, just a little 2dB boost at 1.2K and two very small 1.3dB cuts at 2.7 and 7.4, with narrow Q, and a gentle optical compressor (my LA-2A section in the 610) to bring the spectrum a little more in balance; then used the pultec (and EQP-KT) to boost a little the lower frequency around 100Hz with narrow bandwidth and further little boost at 4K - the Pultec tends to smooth thing even when boosting so it's good for these kind of sounds.

The main mellowing is actually done by filtering the high frequencies with a little bit of delay - 2.2ms, no feedback, "tape" style to avoid to sum an identical waveform. It's a trick that serves you well to smooth the timbre of higher frequencies while keeping things very transparent.

Finally a gentle multiband compressor does the final icing, and an even gentler, very low Q eq brings up the body and brings back a little detail - but nothing more than 0.5dB.

There's a lot more assertive treatment that could have been done, but with this kind of classical stuff I like to keep it as natural as possible, after all it's their timbre you want, not some bigger than life effect :)

And lets I was misunderstood... 7 years old!! Give the boy/girl my congratulations, they can play the violin better than I do!! Lovely to see kids doing that.

As others have said, your original recording is fine, don’t mess with it.
The playing style, and mic position will make more of a constructive difference,
If necessary use a very minimal almost inaudible cut at what you think are the offending frequencies, but honestly, Hugh is right, in the context of what it’s for it’s hardly necessary.

CS70 wrote:

notuno wrote:

CS70 wrote:

notuno wrote:Thanks all for listening to the samples and for the advice. Listening again, I agree that the processed sample lacks clarity.

You can improve the sonics a little bit imho but the room is printed in the recording and there's little to do about that. I did send you a slight worked version yesterday, you should the link in your inbox. The main difficulty was listening to the pitching :D :D but they're students so it's fine :D

Thanks, I missed the message yesterday but I just listened to it now. It's an improvement, did you just cut some high frequencies? This particular fragment is indeed quite out of tune, I should have chosen better...

Two EQs on that one, just a little 2dB boost at 1.2K and two very small 1.3dB cuts at 2.7 and 7.4, with narrow Q, and a gentle optical compressor (my LA-2A section in the 610) to bring the spectrum a little more in balance; then used the pultec (and EQP-KT) to boost a little the lower frequency around 100Hz with narrow bandwidth and further little boost at 4K - the Pultec tends to smooth thing even when boosting so it's good for these kind of sounds.

The main mellowing is actually done by filtering the high frequencies with a little bit of delay - 2.2ms, no feedback, "tape" style to avoid to sum an identical waveform. It's a trick that serves you well to smooth the timbre of higher frequencies while keeping things very transparent.

Finally a gentle multiband compressor does the final icing, and an even gentler, very low Q eq brings up the body and brings back a little detail - but nothing more than 0.5dB.

There's a lot more assertive treatment that could have been done, but with this kind of classical stuff I like to keep it as natural as possible, after all it's their timbre you want, not some bigger than life effect :)

Well, that's way over my knowledge level! One of the reasons why I started tinkering with all this is to start educating myself in postprocessing, so thanks!

CS70 wrote:And lets I was misunderstood... 7 years old!! Give the boy/girl my congratulations, they can play the violin better than I do!! Lovely to see kids doing that.

[/quote]

Will do. This kid can play by memory two full Vivaldi concertos, and other pieces. I can hardly play one movement without memory slips... rather humbling for me. There are other kids on the same level in the school.