I'm looking for recommendations for LAME encoding settings for a whole bunch of recorded radio shows. I used to do a radio show from 1995 - 2003, and the first 6 years or so were recorded directly off the air from an old Panasonic rack system to a VHS hi-fi VCR on the EP speed. VHS! Remember way back then???? Ug.

Anycrap, the broadcasts were actually quite good (most of them), and the signal was strong (most of the time). The VCR was nice for the time and had decent hi-fi sound which I believe added very little distortion or other problems. I'm sampling these via an even nicer semi-pro JVC VCR thru a basic Sound Blaster Audigy sound card, and recording with Adobe Audition in good ol' 16-bit 44.1 kHz. Back in the day, starting sometime around 1998, we used an old program with a Fraunhofer codec at 112kbp CBR, and that was alright for distributing on the web, and they are still quite listenable. BUT, as I get all the older shows archived, I'd like to see if I can improve the quality of the mp3s I will make for the website, while keeping the bitrate around 128kbps.

Radio, of course, is quite limited compared to CD. It has a maximum frequency response of about 15kHz, and has limited signal-to-noise ratio, limited dynamic range, and limited stereo separation. As I understand it, the 112kbps birate like we were using naturally cuts off the frequencies above about 15500 like FM radio, and 128kbps cuts off at higher frequencies. So to make a long question good and long, if I use a VBR around 128kbps (V5?) and also add a lowpass at 15500, would that free up more bits for encoding, and improve the sound compared to V5 without a lowpass? Or is there some more optimal encoding settings altogether?

Best encoding settings for FM radio recordings: Discuss...

Your best bet is to resample to 32khz

V4 or V5 --resample 32

V4 will improve quality and use around 130k due to not much going over 15khz

Regardless of the vbr setting that you use, decreasing the lowpass will not change the quality. The quality is set by the -V value. What it will do is lower the bitrate.

Actually, that's good, since I could go up a notch to maybe V 4, then set a lowpass at 15000, and end up with a smaller file than I would otherwise, right? That's basically what I'm trying to achieve (smallest file that still covers all the sound quality FM is capable of).

Yes. You should find the setting that produces transparent or near transparents results for you (V2, V4, ..) and then add the lowpass or resampling to it.

Don't you mean test for transparency using the lowpass and sampling rate that is the target?

Do not resample. Resampling has many side effects* while the lowpass is the only thing needed.


* Longer pre-echo, LAME's resampler is not bad nor best-of-the-breed, the soundcard will have to resample it up again (especially if using digital out)...

Yeah, I was politely ignoring the suggestion to resample.

I think you guys have helped me narrow down my approach. I'll probably do some A-B-ing with V 3, 4, and 5, while using the lowpass. I'll check a few recordings without the lowpass to see what kind of file size reduction I'm getting.

All my other questions regarding this process would probably go best in another forum here. (Best format for long-term lossless archiving? Good ol' uncompressed .wav? Best method for noise reducing FM recordings on those days when the broadcast sounded like poop? Besides regular mp3s, can I create "enhanced podcasts" for iPods with accompanying still images at timed points? Etc., etc. ...)

Don't use wav for archiving. Apart from being larger, they do not have decent tag capability.

If the FM broadcasts have white noise because of weak signal then you might want to consider converting the files to mono, because most of the noise is identical in the two channels, but with opposite polarity.