sample stereo DTMF tones (2.95 MB)

Interestingly enough, when I tried to encode the above file as a constant bitrate MP3 file (@16kbps & @24kbps), the result was complete silence! Apparently, (I thought) MP3 had some fixed filterbanks that didn't take well to pure tones. MP3 at higher bitrates still sounded weird, and it wasn't until 128kbps that WinAMP actually got it right. When I accidentally dragged my 48kbps file onto the RealPlayer and it played correctly, I realized that the problem was not on the encoding side, but the decoding side: this was a bug in WinAMP. Try listening to the following files in WinAMP, then try listening to them in another program (like RealPlayer or Windows Media).

• 48 kbps MP3 (100 KB)
• 64 kbps MP3 (134 KB)
• 96 kbps MP3 (201 KB)
• 128 kbps MP3 (268 KB)

Lest you all think I'm nuts, or this bug gets fixed, I've provided a WAV output (1.48 MB) of what WinAMP did to the MP3 at 48kbps. This didn't seem to be a problem in other players.

Xing's Variable Bitrate Encoding reproduced the file faithfully at an average of 35.5 kbps at the lowest setting, but could not get any smaller. The VBR file played back just fine under WinAMP, unlike the constant bitrate files. There was little difference (73KB to 90KB) between the highest and the lowest VBR settings for this file:

• Lowest VBR setting (73 KB)
• Highest VBR setting (90 KB)

RealAudio was disappointing, as I could only scale it down to about 20kbps with the free version of their encoder. The test (48 KB) came out a touch scratchy.

The Micrsoft encoder was able to scale to much lower bitrates. When I fed it the tone file and asked it to encode a 5kbps file I was at first surprised to see that the output file contained nothing but silence. Then I realized that all of the frequencies in the file were above 4khz: since the 5kbps encoder was sampling at 8khz, it had missed all of the tones above 4khz! This, incidentally, means that the codec includes a good low-pass filter. Otherwise, there would have been a significant amount of aliasing and I would have ended up with a bunch of noise. Instead, the whole signal was filtered out and I got pure silence.

Listening closely to MS Audio encodings of the file, one can hear the warble and mask when the two frequencies come close together. The two tones are almost fighting each other for dominance. This anomaly does not seem to be as conspicuous at lower sampling rates:

• 20kbps MSA4 encoding, sampled @ 16khz (46 KB)
  (hear the warble & mask)
• 20kbps MSA4 encoding, sampled @ 11khz (46 KB)

High frequency anomalies seem to be a fundamental problem with the MS Audio codec; I will return to this further on in the writeup.