I’d appreciate comments on the sound quality of three audio clips, each of which is 30 seconds long, contained in the following directory:
http://philip.greenspun.com/scratch/audio/
The original file is the .wav. The other two are MP3s recorded at 128 (actually only 64 kbps for a mono file like this) and 64 kbps.
Thanks!
Philip
p.s. None of these audio clips is as funny as the ghetto version of “Who’s on First” that my friend Andrew sent me… http://openpodcast.org/media/2006/2/23/openpodcast_3532.mp3
None of them sound great, but they’re all intelligible.
Answering machine messages aren’t music, and 32kbps mono is usually considered high quality for a digital audiobook.
If I was storing these on a computer for future reference, I would be content with 32kbps mono.
I thought that the MP3 recorded at 128 kbps sounded the best.
Was that clip supposed to be funny?
I have to agree with Paul that the 128 kbps MP3 sounds the best.
I’m with Paul, through my dime-store work headphones, the 128 MP3 is the best. I actually think the wav is a close second, there’s a tinny distortion the mp3s have (the lower bitrate one is way worse) that the wav doesn’t have.
The raw wav is best, of course.
The 128mp3 is only marginally worse: not noticable without a/b comparison.
The 64 is less good. you would miss some more ambiguous single words at this rate. but, only a little more effortful to listen too than the 128 and as intelligible on this recording, esp of just voice.
the emotion comes through crystal clear: she liked the flight, i gather.
Consider Speex or just GSM for voice. GSM is 13kbps IIRC and highly intelligible; Speex is patent-free, has a VBR mode (which makes more sense for voicemail), supports rates from 2.15 to 24.6 kbps, and has the benefit of another ten years of research over GSM.
GSM: http://kbs.cs.tu-berlin.de/~jutta/toast.html
Speex comparisons: http://www.speex.org/comparison.html
It sounds as if she’d like to see you again.
Oh, I liked the 128MP3 best.
Through my laptop speakers there is no difference to take note of. I may say that the recordings were originally recorded at very low rates and we shouldn’t be able to detect very much difference in the first place. However, if you take a sample clip from a Recording Studio and compare different rates then you’ll see a notable difference.