WAV to MP3 conversion -- introductory material needed

My church has decided to make sermons available on the web. They are 
posting versions for high speed and dialup connections(both are mp3). 
They are not particularly happy with the listenability of the low speed 
version. I don't find it too bad, but some on the committee are trained 
musicians.

The compression program was one someone had available. It's quite 
possible the problem lays more in using the software than the quality of 
the software itself. I've volunteered to do the spade work to find out 
what we can expect to accomplish. First I need to be "educated" ;)

A Google search produced too much. Although, www.mp3-converter.com was 
an early hit. I've got 10 pages of it up already and suspect there is 
probably >>> 10 pages more of good info.

1. Can anyone recommend a good _survey_ site?
       For example, one page gives a one sentence description/definition
       of Variable Bitrate Encoding but gives no indication of how many
       mp3 devices can read it.
2. What would be an appropriate newsgroup for asking questions once I've
    read enough to ask them?

Thanks

Richard Owlett wrote:
> My church has decided to make sermons available on the web. They are 
> posting versions for high speed and dialup connections(both are mp3). 
> They are not particularly happy with the listenability of the low speed 
> version. I don't find it too bad, but some on the committee are trained 
> musicians.

What bitrate(s) are you using?  I was working on the exact same problem 
(sermon audio conversion to mp3 for my church) and found 32kbps mono mp3 
encoded with lame to be quite adequate for good voice quality.  If 
someone finds the music intolerable at that bitrate, they should click 
on the high bitrate link.

Also, check out audacity with the "lame" plugin. 
http://audacity.sourceforge.net/
  It is very useful for normalizing  audio levels; as well as removing 
dead air, copyrighted material, and whatever else should be removed.


-- Mark

Mark Borgerding wrote:
> Richard Owlett wrote:
> 
>> My church has decided to make sermons available on the web. They are 
>> posting versions for high speed and dialup connections(both are mp3). 
>> They are not particularly happy with the listenability of the low 
>> speed version. I don't find it too bad, but some on the committee are 
>> trained musicians.
> 
> 
> What bitrate(s) are you using?

I was told last night while other things were going on. I'll have to 
check again.


>  I was working on the exact same problem 
> (sermon audio conversion to mp3 for my church) and found 32kbps mono mp3 
> encoded with lame to be quite adequate for good voice quality.  If 
> someone finds the music intolerable at that bitrate, they should click 
> on the high bitrate link.

An audio nut is supervising it. Music available only on CD ;)
[ Actually more an issue of file size than acceptable quality. ]



> 
> Also, check out audacity with the "lame" plugin. 
> http://audacity.sourceforge.net/
>  It is very useful for normalizing  audio levels; as well as removing 
> dead air, copyrighted material, and whatever else should be removed.
> 
> 
> -- Mark

I don't know what program was used. I've audacity 1.0.0 on my machine 
and am downloading latest as I type.

I'll run some comparisons  of bit rates etc. As I'm interested in speech 
only, would it help If I prefiltered?

On Feb 23, 7:55 am, Richard Owlett <rowl...@atlascomm.net> wrote:
> My church has decided to make sermons available on the web. They are
> posting versions for high speed and dialup connections(both are mp3).
> They are not particularly happy with the listenability of the low speed
> version. I don't find it too bad, but some on the committee are trained
> musicians.
>
> The compression program was one someone had available. It's quite
> possible the problem lays more in using the software than the quality of
> the software itself. I've volunteered to do the spade work to find out
> what we can expect to accomplish. First I need to be "educated" ;)
>
> A Google search produced too much. Although,www.mp3-converter.comwas
> an early hit. I've got 10 pages of it up already and suspect there is
> probably >>> 10 pages more of good info.
>
> 1. Can anyone recommend a good _survey_ site?
>        For example, one page gives a one sentence
> description/definition
>        of Variable Bitrate Encoding but gives no indication of
> how many
>        mp3 devices can read it.

mp3 (MPEG-1 Audio Layer 3) is a decoding standard.  You can
encode it any way you like; but all compliant decoders are
supposed to decode it exactly the same way (any differences
you hear will more likely be in the sample rate conversion,
filtering, equalization, volume control and/or AGC stuff
in the playback path following the actual mp3 decoder).

For speech, I'd give VBR a try.  The poetic pauses in
speeches and sermons don't need very many encoding bits,
which can then better be used for the following consonants.
You might also want to make sure the miking and recording
is as clean as possible before attempts at compressing.



IMHO. YMMV.
--
rhn A.T nicholson d.0.t C-o-M

Richard Owlett wrote:
> Mark Borgerding wrote:
> 
>> Richard Owlett wrote:
>>
>>> My church has decided to make sermons available on the web. They are 
>>> posting versions for high speed and dialup connections(both are mp3). 
>>> They are not particularly happy with the listenability of the low 
>>> speed version. I don't find it too bad, but some on the committee are 
>>> trained musicians.
>>
>> What bitrate(s) are you using?
> 
> I was told last night while other things were going on. I'll have to 
> check again.

The "wide bandwidth" feed was encoded at 32 kbps, "dial up bandwidth" at 
8 kbps. I find the later to have acceptable quality for sermons.


> 
> 
>>  I was working on the exact same problem (sermon audio conversion to 
>> mp3 for my church) and found 32kbps mono mp3 encoded with lame to be 
>> quite adequate for good voice quality.  If someone finds the music 
>> intolerable at that bitrate, they should click on the high bitrate link.
> 
> 
> An audio nut is supervising it. Music available only on CD ;)
> [ Actually more an issue of file size than acceptable quality. ]
> 
> 
> 
>>
>> Also, check out audacity with the "lame" plugin. 
>> http://audacity.sourceforge.net/
>>  It is very useful for normalizing  audio levels; as well as removing 
>> dead air, copyrighted material, and whatever else should be removed.
>>
>>
>> -- Mark
> 
> 
> I don't know what program was used. I've audacity 1.0.0 on my machine 
> and am downloading latest as I type.
> 
> I'll run some comparisons  of bit rates etc. As I'm interested in speech 
> only, would it help If I prefiltered?

Ron N. wrote:

> On Feb 23, 7:55 am, Richard Owlett <rowl...@atlascomm.net> wrote:
> 
>>My church has decided to make sermons available on the web. They are
>>posting versions for high speed and dialup connections(both are mp3).
>>They are not particularly happy with the listenability of the low speed
>>version. I don't find it too bad, but some on the committee are trained
>>musicians.
>>
>>The compression program was one someone had available. It's quite
>>possible the problem lays more in using the software than the quality of
>>the software itself. I've volunteered to do the spade work to find out
>>what we can expect to accomplish. First I need to be "educated" ;)
>>
>>A Google search produced too much. Although,www.mp3-converter.comwas
>>an early hit. I've got 10 pages of it up already and suspect there is
>>probably >>> 10 pages more of good info.
>>
>>1. Can anyone recommend a good _survey_ site?
>>       For example, one page gives a one sentence
>>description/definition
>>       of Variable Bitrate Encoding but gives no indication of
>>how many
>>       mp3 devices can read it.
> 
> 
> mp3 (MPEG-1 Audio Layer 3) is a decoding standard.  You can
> encode it any way you like; but all compliant decoders are
> supposed to decode it exactly the same way (any differences
> you hear will more likely be in the sample rate conversion,
> filtering, equalization, volume control and/or AGC stuff
> in the playback path following the actual mp3 decoder).
> 
> For speech, I'd give VBR a try.  The poetic pauses in
> speeches and sermons don't need very many encoding bits,
> which can then better be used for the following consonants.
> You might also want to make sure the miking and recording
> is as clean as possible before attempts at compressing.
> 
> 
> 
> IMHO. YMMV.
> --
> rhn A.T nicholson d.0.t C-o-M


I'm not concerned with the quality of miking. We have been actively 
analog recording services, including music, for about ten years. With 
several trained musicians (and a couple teaching at college level) there 
have been educated ears listening to the result ;)

I'll try the VBR on our "high speed" version (encoded at 32 kbps). Our 
"dial up" version is encoded at 8 kbps.

With the 8 kbps encoded version, would there be any file size advantage 
to band pass filtering the original before encoding? I'm thinking about 
some of the description of the encoding theory I read on how some 
frequencies are more "important" depending on relative intensity. I can 
say what is "important" (a particular speaker), and not leave the 
"decision" to software which tries to do general optimization.

Richard Owlett wrote:
> I'll try the VBR on our "high speed" version (encoded at 32 kbps). Our
> "dial up" version is encoded at 8 kbps.

8 kbps? MP3 i.e. MPEG 1 Layer 3 starts at 32 kbps. 8 kbps is possible
with MPEG 2 Layer 3, using half the sample-rate of MPEG 1. Anyway, I
wouldn't be surprised that audio encoded with the lowest-possible
bitrate setting sounds poor. I would have expected the "high-speed"
version around 128 kbps, and the "low-speed" version at 32 kbps. Not
sure about your country, but in mine, modems manage to receive 32 kbps
most of the time.

> With the 8 kbps encoded version, would there be any file size advantage
> to band pass filtering the original before encoding? I'm thinking about
> some of the description of the encoding theory I read on how some
> frequencies are more "important" depending on relative intensity.

An MPEG audio encoder usually has built-in band-limiting filters; if the
bitrate is set too low, it won't even look at the high frequencies (I'm
not sure, maybe this is even required by the standard).

If you're transmitting just speech, using a specialized speech codec
(e.g. Speex) might improve quality. But that might be harder to set up
for encoders and listeners, as it isn't as ubiquitous as MP3.


  Stefan

Stefan Reuther wrote:
> Richard Owlett wrote:
> 
>>I'll try the VBR on our "high speed" version (encoded at 32 kbps). Our
>>"dial up" version is encoded at 8 kbps.
> 
> 
> 8 kbps? MP3 i.e. MPEG 1 Layer 3 starts at 32 kbps. 8 kbps is possible
> with MPEG 2 Layer 3, using half the sample-rate of MPEG 1. Anyway, I
> wouldn't be surprised that audio encoded with the lowest-possible
> bitrate setting sounds poor. I would have expected the "high-speed"
> version around 128 kbps, and the "low-speed" version at 32 kbps. Not
> sure about your country, but in mine, modems manage to receive 32 kbps
> most of the time.
> 
> 
>>With the 8 kbps encoded version, would there be any file size advantage
>>to band pass filtering the original before encoding? I'm thinking about
>>some of the description of the encoding theory I read on how some
>>frequencies are more "important" depending on relative intensity.
> 
> 
> An MPEG audio encoder usually has built-in band-limiting filters; if the
> bitrate is set too low, it won't even look at the high frequencies (I'm
> not sure, maybe this is even required by the standard).
> 
> If you're transmitting just speech, using a specialized speech codec
> (e.g. Speex) might improve quality. But that might be harder to set up
> for encoders and listeners, as it isn't as ubiquitous as MP3.
> 
> 
>   Stefan
> 

To paraphrase Will Rogers, "All I know is what I see on my monitor."

Using Windows Explorer, when the cursor hovers over the "dial up" 
version of the file it is identified as 8 kbps. When it hovers over the 
"high speed" version it indicates 67 kbps encoding.

Common "dial up" modems are rated at 56 kbps. Windows typically says I'm 
connected at 44 kbps. Throughput tends to run around 2.5 kBps.

For my current encoding experiments, I am using  the free version of 
WavePad v 3.05.

The perceived quality of the 8 kbps seems to depend heavily on the 
specific encoder and settings used. Following Ron Nicholson's suggestion 
I am experimenting with Variable Bit Rate encoding. Encoding at 32->8 
kbps gives me as satisfactory audio quality as the 32 kbps constant rate 
with approximately 1/2 the file size of the 8 kbps.

There may be more that can be done. As I don't yet have a copy of the 
_ORIGINAL_ wav file, my experiments have used a wav file _recreated_ 
from the 32 kbps. For each objection to that experimental procedure I 
could probably add 2 or 3 more. Ya work with what ya got ;)