[asterisk-users] drop dead fix

[Steve Edwards]

> [mailto:asterisk-users-bounces at lists.digium.com] On Behalf Of Kevin P. 
> Fleming
>
> There were some comments in other replies about your files being 'quiet' 
> (low average volume level)... this won't help your situation at all, 
> because it means that any artifacts caused by resampling and 
> compression/decompression will end up at a relatively high amplitude 
> compared to the original signal (resulting in a low signal-to-noise 
> ratio), and when the listener increases the volume level on their 
> listening device, the noise level will be increased along with it. For 
> these sorts of tasks, you really do want the source material recorded at 
> a fairly high volume level.
>
On Fri, 15 Oct 2010, Danny Nicholas wrote:

> This appears to be the resolution to my problem -
>
> #1. Get my "recording talent" in an isolated environment so I can get 
> "clean, loud" recordings
>
> #2. Dump the Audacity and Audiologic steps and just use SOX with the 
> highpass and lowpass filters.

1) firstmenu.wav.wav is recorded so low, it just looks like line noise in 
Audacity. Unless you can re-record at a reasonable level, you're always 
going to be fighting this "sow's ear."

2) I use "normalize" (http://normalize.nongnu.org/) to normalize from the 
command line, but it does not deal with DC offset like Audacity will. 
Eliminating your DC offset issue should also be a goal of improving your 
recording environment. Newer (than provided with CentOS 5.5) versions of 
sox can do "dcshift."

3) Stick with ULAW or PCM (wav). You only have to be concerned with 
supplying audio encoded appropriately for the first hop in your delivery 
path.

-- 
Thanks in advance,
-------------------------------------------------------------------------
Steve Edwards       sedwards at sedwards.com      Voice: +1-760-468-3867 PST
Newline                                              Fax: +1-760-731-3000