I have an mp3 128 Kbps music file with particular parts where a series of single note only on a cello is played initiated with a bell click. Less than a second after a bell click and cello note playing starts a very faint Morse code like series of bursts in 2 - 5 kHz range for a couple of seconds then fades away.

I was able to localize their frequency range and duration with iZotope RX on a time-frequency spectrum.

They happen only in the parts of a cello note playing after a bell click in some parts even fainter than in the others.

Could they be attributed to errors in mp3 compression or to other factors?

  • 1
    Can you post a sample of the audio? – AJ Henderson Feb 8 '14 at 18:40
  • 1
    It would help if you attached the file. – Sergey Stadnik Feb 11 '14 at 0:46
  • Could it be just slight distortion from overdriving the channel? The bell could "trigger" the clipping. – Tobias Schmidt Feb 12 '14 at 22:57

Yes - most definitely this is to do with mp3 compression. Strings in particular require a much higher bitrate to be reproduced through mp3 without artifacts. You're not going to get much mileage out of a strings recording under 192kb/s. Adding a bell is just asking for trouble. The way that the mp3 encoding works will mean that the available bit rate is taken up encoding the bell sounds which are of a particular set of harmonics while the remaining bits in the stream probably aren't enough to handle the cello - which has an immensely complex harmonic structure. This sounds very much like a test case recording. Is there a copy of the recording available?


"Morse code like series of bursts in 2-5kHz range for a couple of seconds" very much sounds like RF interference from a mobile phone (those tend to occur before a call/SMS or when doing a call, and in sparse patterns in between). Now you associate this sound not with just a single passage but rather a recurring combination of elements. That could be

a) because a sound sample is being used repeatedly b) a mobile phone (ring tone?) is involved in the sound production. This would be rather hard to time so it's unlikely unless we are talking about a combination of a) and b): a ring tone sampling being used.

Reuse of a sample disturbed by RF interference sounds to me more plausible than compression artifacts of similar (rather than identical) music resulting in an audibly unique pattern like the one you describe.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.