Mains hum: notch filter alternative?

I’m not deeply experienced with audacity and hope for some help to solve a hum problem when recording from a turntable.

The hum noise is inducted from a transfomer of the turntable, 50 Hz here, and has about ca -58 dB only.

It grows +3 dB when the tonearm comes to the middle of the turntable, an it is stronger at one channel (what may depend on the coils direction of the used cartridge).

A notch filter would work, but I’m afraid about the changed phase at the surrounding frequencies.

I would prefer adding a 50 Hz Signal with 180° different phase to only one of the stereo-channels, if that helps. Similar with 150 Hz if necessary.

How can I adjust an additional track with 50Hz at -60dB exactly to that 180° and mix it to one channel? Or can you help me with another advise?


The likely problem will be if the recorded hum is not exactly 50 Hz, but say 49.99 Hz or 50.01 Hz as this will cause the relative phase between the original hum and the added tone to drift.

Another likely problem is the difficulty of accurately matching the amplitude of the added tone to the hum.

To generate a 50 Hz tone at -60 dB, with a starting phase of 180 degrees, you can use the Nyquist Prompt

;version 4

(mult  (db-to-linear -60)
      (osc  (hz-to-step 50.0) 1 *sine-table* 180.0))

You can try the built-in Effect > Hum Remover.

Screen Shot 2020-01-01 at 8.02.11.png

If you follow the General Recommendations, you will be exporting your raw captures as WAV, hum and all, so you will be able to continue your recordings and try different fixes later. Audacity Projects are not recommended for this.


A notch filter would work, but I’m afraid about the changed phase at the surrounding frequencies.

You can’t hear a phase shift without a reference. For example, if you phase-shift the left channel without changing the right you can get some “weirdness”. And, there is already a ship-load of phase shift in the RIAA recording & playback equalization and record cutting & playback process.

This is where you separate out each hum tone and harmonic overtone, and Nyquist your brains out to derive the exact anti-hum tone and then apply it in multiple passes to make sure it did a good job.

How close are you to retirement? That will be a career move.

I don’t see this mentioned explicitly in the discussion so far. Hum isn’t singular. It’s plural. Nobody can hear 50Hz, but everybody can hear the odd and even multiples caused by power line slop and motor distortion. That’s why Effect > Hum Remover has those Odd and Even sliders.

This is a very bad but very real example of hum in the US.

Screen Shot 2020-01-01 at 9.08.39.png
Count the spikes. The far left and largest one is 60Hz as expected. Next up is 180, then 300, 420, 540, 660…

You’ll need a cancellation tone for each of those plus the sampling and phasing dance.


That looks very nice with the prompt.

Yes. I hoped that the provider would adjust the frequency very soon, but as I just could read that corrections may come every hour or less and the frequency vary more than I thought. Now I guess I need another method.

Sorry but I’m not sure that I got it- if I reduce hum by the use of Audacity it should not be recommended or not be the purpose of Audacity?
Or just that I should do it in a different step from recording?

I would run filter on both channels. I believe that some phase effects can be heard: at higher frequencies as different depth of the virtual room, and as slowness at lower frequencies like a fourth order enclosure (vented box).

Yes thank you, the treshold means a limit and loud areas keep untouched?

Now I have the Idea of combining a notch filter at 50 Hz with a little rise at 50 Hz by equaliser to reduce the effect oft the filter.

But would that produce more of distortion oder noise (something like math rounding) at each filtering process?


This problem just screams the connections on the back of the cartridge are broken or loose. This is a decades old problem with turntables and even the most inexpensive ones have shields and connections to avoid it.

Sound on a record isn’t “normal.” Big powerful low tones are suppressed so they fit in the groove and high tones are boosted to help with surface noise. When you play the record, the RIAA filter has to put everything back where it was and that means boosting low tones, exactly where hum lives.

The turntable is plugged into a preamp or music system, right? Did you remember to connect that thin black ground wire between the turntable and preamp? There are three turntable connections, not two.

Screen Shot 2020-01-01 at 9.56.42.png

Even the tiny phono systems such as the Behringer UFO-202 have a place to screw the third ground wire (on the left).

Screen Shot 2020-01-01 at 10.02.04.png

Sorry but I’m not sure that I got it- if I reduce hum by the use of Audacity it should not be recommended or not be the purpose of Audacity?

Or just that I should do it in a different step from recording?

It’s a very New User mistake to capture work, whether voice, live music or records and then edit, filter, change and patch that one sound capture until it’s perfect. If anything happens to Audacity or the computer in the middle, there is no backup. The sound data goes away and you will be in the collection looking for that record to transfer it again.

Also as a secondary consideration, vinyl records gently but permanently degrade at each playing. That and you should wait a bit before a second playing to give the vinyl a chance to rest. I’m not making that up.

Much better to play the work once—errors and all—and Export a perfect quality WAV sound file and work on a copy of that WAV file.


I still believe that the transformer induces into the coil. That with different level left compared to right because the 540ML has Coils in V-Constellation, and the more the nearer the cartridge comes while playing.

I’m not sure about your information.

The hum from my recordings has a low level, and I believe that caring about 50 Hz and perhaps 150 Hz is enough in my case.

Some kind of notch filter at that frequencies with -18 dB to -12 dB might be enough.

The Hum Remover has more advantages against the Notch Filter and uses an anti-phased signal?

Found it, the Hum Remover uses Notch Filter, but depending on level:

This effect therefore has a threshold level control so that only quiet sounds (where the hum will be most noticeable) are filtered.

The default 1 odd and default 0 even harmonics would work for me, but I am not sure about the treshold and the intensity of the filters. I will try a notch filter like 50 Hz and a Q of 9 and an additional small peak of +3 or +4 dB by Equalizer to reduce the notch.

I have a very strong bias to fixing the problem rather than trying to patch the sound damage later.

You can get Mu Metal sheets to shield against magnetic interference. In some cases, plain steel may work enough.

These people have explanations for shielding applications.

Many (all) legacy microphone transformers have Mu-Metal shielding to keep random magnetic fields from affecting the sound.

Screen Shot 2020-01-02 at 16.12.13.png
You can’t shield the cartridge or change the arm, but if you have access to the transformer underneath, you may be able to help there. Magnetic shields do not have to be grounded. Just putting a plate of steel or MuMetal between the transformer and the arm may do it.


Thank you for the idea, I tried steel with no real success already. Mu Metal would be expensive. Could a copper foil around -like a closed ring- help?

The hum is not extremly disturbing. But the speaker I am using at the moment are not the best, so that I dont know how bad it is. And it is louder at one channel, where the attached pictures may show the sum of both channels only.

I found other unwanted effects when recording from turntable. There is noise at 7 Hz, tonearm resonance, and noise around 29 Hz - if that is not from the tonearm too, it could be the motor or it might come from the record.

I want to use a Low Pass Filter at 14 Hz with 6 dB or/and a Notch Filter of 7 Hz and Q 4.

Should I use a notches at 29 Hz and at 50 Hz?

Could a copper foil around -like a closed ring- help?

It’s less likely. That’s more a Faraday Shield against electro-static effects. You need Magnetic Mud and that’s where MuMetal comes in. It’s expensive and even better, delicate. They warn you not to hammer it or cause physical stress or it will lose its properties.

However, copper sheet is (relatively) cheap and worth testing.

I don’t entirely agree with you the Left pickup is closer and therefore louder. It’s much more likely the Left pickup is aligned better with the hum field. As you point out, the two pickups inside the cartridge are at angles to each other. Orientation in a hum field is important.

Can you move the transformer? It wouldn’t take much. Hum effects decline as the square root (or square) of the doobly-do and a couple of inches should do it. How are your soldering skills?

Obsessive Engineer will, of course, want you to create a custom bucking hum field to cancel the effects.


Could be the best solution. I’ not experienced with the newer lead-free solder, and would need good glasses. And a lot of patience.

My old demagnitizer coil helps about 3 to 6 dB, but becomes hot after some time…

But what could be the reason for that 29 Hz noise; tonearm, bearing, or some kind of ‘direct drive stepping’?

reason for that 29 Hz noise

Is the turntable multi-speed? Do both speeds do the same thing?

I’m trying to think of a good way to actively search for the hum field.

I went hum searching with a portable microphone system when I got tired of managing a very quiet hum in my studio. I was outfitted like one of those guys with metal detectors at the beach except a microphone at the end of a stick instead of the interference coils. I found an amplified bass cabinet which produced very low volume hum in the room even though it was turned off. Pull the power cable. Boop. Problem gone.

We may find out why Belt-Drive turntables put the motors far away from the pickup.

And just to cover the obvious, does the hum stay constant if you slide the turntable left or right? Can you plug in the turntable power “backwards”—does it have a non-polarized cord?

It is important to observe a change, good or bad. If you can make it worse, that’s extraordinarily good to know.


non-polarized cord

Never mind. If you’re in a 50Hz country, it’s a good bet you have on of those big, protected plugs with the fuses in it.


At the moment with a little help of the demagnitizer coil behind the turntable the mains hum is not very disturbing, and I could lower the other mud by the use of to different mats a little. A complete “cleaning” with a strong subsonic filter will lower a lot of bass too, so I want to be careful there.

I am not satisfied with the sound of the VM 540 ML at the moment. Now I want to try RIAA filtering by audacity and just visited, but could not find an Equalizer Preset for that purpose.

Do I need a complex filter like ‘linear phase’ or FIR? Just a simple low pass 6 dB from 10 Hz would not fit good enough - it should be two functions around 1000 Hz like 6 dB filters with 3 dB at 500 and 2122 Hz?

I just found the RIAA Equalizer-Curve in Audacity submenu.

The sound is rougher than from the preamplifier, much better.
Still a little braking anywhere although hi-hats become scratchy, but it is more realistic.

The phase-shifts are the same at hardware or Audacity-filter?

A notch filter is better for removing / reducing hum than a strong subsonic filter.

The first problem with using a subsonic filter is that musical bass frequencies frequently go lower than 50 Hz.
The next problem is that to create a substantial reduction at 50 Hz, while not reducing frequencies too much in the 60 to 100 Hz range, you would need to use a steep cut-off filter, which will introduce ringing close to the cut-off frequency.
The next problem is that while FFT based FIR filters have linear phase, they are not well suited to low frequency applications because they require an extremely large FFT size, but on the other hand, while IIR filters can produce a steep low frequency cut-off, they don’t have linear phase.

Whatever type of filter you opt for, there are always compromises, and in the case of removing / reducing a 50 Hz hum, a notch filter is likely to be the best compromise.

Notch filters also have problems with ringing, though this can be mitigated by mixing a portion of the unfiltered sound back in with the filtered sound. For example, a 50:50 mix of notch filtered sound with unfiltered sound will give you 6dB attenuation at the notch frequency, and only half the amount of ringing.
You can also adjust the Q of the notch filter to minimise unwanted effects. For example, for music that does not have much very low bass, you may decide to use a low Q setting, which will give a broader notch, but less ringing. For music that has a lot of low bass, you could try a high Q setting so as to make the notch narrower and check that ringing artefacts are not audible - reduce the Q if ringing is noticeable.

As a ballpark figure for the Q of a notch filter, try around “2” and adjust to taste.