Home » Uncategorized » Two Important New Papers from Henrik

Two Important New Papers from Henrik

Follow World Hum Map and Database Project on WordPress.com
Follow World Hum Map and Database Project on WordPress.com

Henrik, one of our “scientists in residence” has written a long-awaited paper on tracking down external environmental noise using very basic tools. I will feature prominently on this blog and on the Hum Map website for the near future. I hope that this paper will not only help create a purer dataset for our project but will aid people who are suffering from nuisance, low-frequency noise that arises from classic anthropogenic sources.

The paper is written in two parts.

Read Part One here.

Read Part Two here.


  1. Kurt says:

    I wonder if the iphone/ipad app: FFT Plot can be used for this purpose

  2. Henrik says:

    Maybe someone who has done this with iPhones can answer this specific question.

    What you should look for is a real-time audio analyzer with FFT (Fast Fourier Transform) functionality. We are searching for low frequencies, so we want a low sampling rate if possible, since we need a high frequency resolution, at least down to below 3 Hz. The frequency resolution (or Hz/bin) is the sampling rate divided by the FFT size. If the app has only 44.1 kHz or higher, we need an FFT size of 16384 to reach even 2.7 Hz resolution. This will drain the battery faster, since the processor load will be high. That’s why the Spectroid is good, since it has sampling rates down to 8000 Hz. I used 11025 Hz in the instruction to create a best match with the TASCAM recordings, but for just the visualization on the phone, any sampling rate is OK as long as we reach a good enough frequency resolution.

    It is also not ideal to use extreme resolution, since out of the man-made sounds, only the power line harmonics are absolutely stable. All machine sounds tend to “wobble” a little, so a resolution below 1Hz is no longer useful.

  3. Brenda says:

    I’m quite surprised you don’t mention anywhere the difference between dB(A) dB(C) and d(Z), although this is an essential distinction in measuring low-frequency and infrasound. By many governmental agencies dB(A) is used due to following the ISO-guidelines (which are based on A-weighting) when measuring low-frequency sound, which leads to false conclusions, since A-weighting downgrades the level of low frequency sound, based on the ASSUMPTION that EVERY INDIVIDUAL is equally less sensitive to low-frequency sound.
    Applying the scale and methods mentioned in this document leads to false conclusions about the Hum being internally generated or coming from an external source. Application of the wrong weighting by some authorities and researchers is actually one of the big causes of frustration for Hum-sufferers who are trying to get recognition for their suffering. Luckily an increasing number of researchers and institutions are aware if this issue and apply C- or Z-weighting in evaluating low frequency sound levels.
    I suggest you take the document down and add information about the difference between A-weighting, C-weighting and Z-weighting.
    Furthermore, many (if not most) people experience a hum below 70Hz, which makes most smartphones useless as a sound measuring device due to the lowpass filter, which you mention in your document.

    • With all respect, I think you need to bring yourself up to speed on the work that attends this project. The issues you raise have been explored in depth on this blog. We are well aware of the various weighting scales. I would refer you to Bernie or Henrik if you want to pursue the finer details of that. Regarding cell phone recordings of low-frequency noise, I have been surprised recently at the quality of some recordings that have been sent to us. Of course, a more professional setup is always preferred.

    • Henrik says:

      These papers were not intended to be a textbook on audio mesasurements. What we try to do is to identify exact frequencies of individual tones (for possible identification of the source), and their approximate levels (for comparison to the human hearing threshold). Weighting curves relate to the concept of loudness, where all audible sounds are added together. I agree with you that all authorities use A-weighting to show as low noise levels a spossible, but that is a different purpose. For our purposes the described methods are adequate, including the use of smartphones.

    • The technical terms “decibels” and “Frequency Weighting” as in dB(A), dB(C), and dB(Z) {presumably this is what Brenda meant by d(Z)} do not apply to the Hum I experience which is unequivocally of internal origin and NOT a “real” acoustic signal. It can NOT be displayed or recorded by any usual equipment/procedures.
      http://electronotes.netfirms.com/ENWN41.pdf http://electronotes.netfirms.com/ENWN53.pdf

      I can not emphasize too strongly how much my Hum is a very real thing, but is INTERNALLY generated, on all evidence, especially the “head shake interruption”. I strongly suspect many others hear their Hum as being equally as compelling a phenomenon – almost a “personality”. This is potentially why a good number of Hum hearers insist that there are no valid recordings (nothing like our own “old friend”).

      Unquestionably, likely as many people as those on this site who hear “the Hum” (internal), instead hear “a hum” from perhaps an engine, a transformer, or some infrastructure. Such a true acoustic sound, potentially identifiable if one ties harder, could be displayed/recorded. Given that the usual pitches, of Hums and hums, are low, as are the loudness levels, recording possibilities begin at an (unfamiliar) already difficult level and get harder, as discussed in detail here:

      Except as a scientific demonstration, we should be forgiven for not investing time/money in trying to record an acoustic signal we strongly suspect does not even exist physically. Such a recording attempt would fail and prove little or nothing since the cause of the failure would not likely be evident. This is especially true given the (seldom appreciated) difficulties I identified in the webnote-54.

      But if the evidence points to an external source, attempting a recording, possibly looking for a mechanical source, is appropriate, and Brenda is correct to suggest that A-weighting should not be used. For example, the 64 Hz I hear would start out 25 db down! Use Z (unweighted) if you have it, otherwise C is not so bad. Below is a note on weighting I wrote for some local legislators considering a noise ordinance. I kept it simple! Of course, they never read it!


    • Henrik says:

      This discussion thread illustrates what is alreday said on the front page of thehum.info: Millions of people around the globe suffer from persistent low-frequency noise. An estimated 50% of these are caused by actual environmental noises or vibrations, and the other half are internally generated hum perceptions, the mechanisms of which are the subject of our research. Only environmental noises can be recorded.

      I have presented the tools for distinguishing the two sources from each other, and to trace and possibly remedy the external sources. The real research continues on the internal sources, and the most important information here is to submit an accaurate frequency match and an accurate description of the characteristics of the hum when you report your observations.

  4. Henrik says:

    A few hints for identifying hum componets in recordings or in the Spectroid screen:

    If you see stable tones between 45 and 49Hz (US/Canada 54-59Hz), the most likely source is an electric motor. The bigger the motor, usually the closer to the line frequency.

    Noises that visibly change in volume and frequency are usually traffic noises.

    If you see exactly 90Hz, it may be a 5400 rpm hard disk drive. Switch off all computers and recording devices using HDDs when chasing the Hum! Their cooling fans also produce a wide spectrum of low-frequency noises.

    if you see exactly 100Hz (US/Canada 120Hz), the likely source is a transformer or contactor/switch. A stable 120Hz tone may also come from a 7200 rpm hard disk drive.

    If you see exactly the line frequency (+-0.5Hz), your microphone, the microphone cable or the recorder circuitry is picking up magnetic fields, not sounds. Only electret microphones should be used for hum-hunting, and any recording device should be battery-operated. The built-in microphones in smartphones and tablets are also very immune to magnetic fields.

    Avoid using a laptop’s built-in microphones. They are usually heavily filtered, and always pick up the sound of the the cooling fan, and the hard disk, unless you have an SSD.

    For determining exact frequencies, use 1.3Hz resolution and Rectangular Window in the Spectroid or other FFT app.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: