[geeks] Audio Recording

Dan Sikorski me at dansikorski.com
Tue Jan 25 08:18:02 CST 2011


On 1/24/2011 4:24 PM, Phil Stracchino wrote:
> On 01/24/11 15:15, gsm at mendelson.com wrote:
>> The problem is that PC sound cards stink. They are built for playing sound to
>> small speakers, chatting with $1 condenser microphones and recording old
>> tapes to low bitrate MP3s.
> The problem is that *the majority of inexpensive PC sound cards and
> onboard sound chipsets* stink.  There are some extremely good ones out
> there.  The EMU10K/EMU10K1 chipset is one to look for; it's capable of
> 96KHz, 24-bit audio.
>

Regardless of the quality of the chipset, the inside of a PC can have a 
ton of interference that will affect the analog circuitry.  For example, 
my Dell Precision M65 gets a bunch of interference from the IR port in 
the headphone output.  (The IR port is disabled for this reason.)  
Listen for ticks and hiss in the recordings, you might be fine or it 
might be a big problem, every PC is different.  Personally, I'd probably 
use the SGIs to get the material from the source, then a PC for 
compression, but i don't have much experience with SGI audio.

>> I think the only compression available for general use is MP3. Most portable
>> devices only decode MP3 files, although there are plenty that decode the
>> MP4 audio compression, also known as AAC.
> It should be noted that a good MP3 encoder such as LAME, with
> appropriate settings, can be very good indeed, at the cost of a larger
> usage of disk space, while still producing output several times smaller
> than the raw PCM audio files.

I'm assuming that the live performances mentioned in the original post 
are music, and not spoken word or anything else.  I would go for a 
lossless codec (FLAC) regardless of how good you think they sound now.  
Even if you are the only one who will ever listen to the recordings, in 
10 years you may have significantly better playback hardware and then 
notice the compression artifacts that you couldn't discern on your older 
gear.  If your source material isn't really high quality, it's defects 
will only be compounded by lossy compression, so don't just write it off 
saying that it's low quality anyway and doesn't matter.  You may not 
need to go crazy with the bit depth and sampling rate, just don't cut 
yourself short, unless you are looking at a huge amount of material, 
storage is relatively cheap.

For an extreme example, i had a coworker who, years ago, ripped his 
entire cd collection on his laptop and encoded them with 96kbps mp3 
because he couldn't tell the difference on his crappy built in laptop 
speakers between that and the original CD.  As soon as he listened to 
those files on something else, he immediately figured out that he had a 
problem.

     -Dan


More information about the geeks mailing list