[App_rpt-users] Female voice announce levels

[Ken]

>From: app_rpt-users-bounces at qrvc.com
[mailto:app_rpt-users-bounces at qrvc.com] On Behalf Of Ken Leisten
>Sent: Friday, December 10, 2010 11:38 AM
>To: Tim Sawyer
>Cc: app rpt-users
>Subject: Re: [App_rpt-users] Female voice announce levels

>If you read the instructions on the support pages for creating custom voice
announcements with the Audacity program, you'll see >that you are instructed
to set the final level of the voice file to -6 db before saving it. I assume
that this is based on 5Khz >deviation radios. 
 = = = = 

This is not right - if you set up for 2.5 kHz deviation, then the loudest
sounds will be 2.5 kHz deviation no matter what (assuming you have selected
the "limit" option in the config file). As someone else pointed out, that
means running the test tone so it gives 1.5 kHz deviation in RADIO TUNE
TXVOICE (giving you 2.5 kHz at full blast).

The -6dB guideline is just to allow some "headroom".  If you have two
sources with 0.5 (relative voltage) level, the waveform at times will add up
to just 1.0 voltage level and you will stay clear of the limiter.  If you
have more signals, or higher levels, then you will start hitting the output
limiter and things will sound compressed. Without the limiter you will get
nasty distortion.  

Another factor (noted in the instructions) is that if you are using
composite audio, the levels for tone and voice must not exceed 1000 in total
- or you will get distorted audio, even *with* the limiter enabled.

Having said that, the pre-recorded sounds are actually hotter than this
guideline. "connected.gsm", for example, hits -3 dB in Audacity as do
"version.gsm" and others.

The recordings are loud and slow so that you will understand them. But many
people would prefer quiet fast ones most of the time.  If some person were
to record and post an alternative library of the letters, numbers and a few
choice words, that would be a useful resource to the user community.

Ken
KE2N