Quality of LAME encoded MP3?

Archived from groups: rec.audio.pro (More info?)

I have a Wave recording of someone speaking faintly, which I wish to
transfer to MP3 for space and compatibility reasons. The option which
has been suggested is encoding by LAME, which I see can encode at
320kbits/second. This gives me a recording which to my untrained ear
sounds almost as good as the original Wave, and is only 25% the size.

I don't trust my untrained ear, hence the question to this group. Is
max quality LAME encoded MP3 really as good (or almost) as source
Wave, for quiet speech? The original recording was on Sony minidisc
which uses ATRAC compression (then taken out over optical link), so
we're not talking pristine audio quality to start off with.

Also, the original recording was on a mono microphone, but for some
reason the Wave is stereo. Using RazorLAME, what should I set the mode
to; stereo, joint stereo, forced joint stereo, or mono? I am guessing
mono, because the signal should not have any stereo separation, having
originally been mono.

Archived from groups: rec.audio.pro (More info?)

<maxi@falkirkreserve-nospam.com> wrote in message
news:fi7fi0tk00s932ghn41kom1ln5ejulsljc@4ax.com...
> I have a Wave recording of someone speaking faintly, which I wish to
> transfer to MP3 for space and compatibility reasons. The option which
> has been suggested is encoding by LAME, which I see can encode at
> 320kbits/second. This gives me a recording which to my untrained ear
> sounds almost as good as the original Wave, and is only 25% the size.
>

The archive quality preset on LAME is almost indistinguishable from the
original file to me.

> I don't trust my untrained ear

You should.

Archived from groups: rec.audio.pro (More info?)

"Sugarite" <nobody@home.com> wrote in message
news:lO7Wc.9395$0c.4709@read1.cgocable.net...
>
> Encode at 128kbps mono LAME MP3. That will be the same quality as a
stereo
> 256kbps MP3, which is the optimum bitrate for MP3. 320kbps actually
sounds
> worse. Don't ask me why, I've heard the difference and seen some
compelling
> evidence via sonograms.
>

I prefer the lower (128k) to 320k for listening in the car. It's not as
"true" as the 320 but works much like compression for bringing things
forward.

Archived from groups: rec.audio.pro (More info?)

<maxi@falkirkreserve-nospam.com> wrote in message...

> Also, the original recording was on a mono microphone, but for some
> reason the Wave is stereo. Using RazorLAME, what should I set the mode
> to; stereo, joint stereo, forced joint stereo, or mono? I am guessing
> mono, because the signal should not have any stereo separation, having
> originally been mono.

Upon recommendation of folks here, I've put LAME into a couple of PCs.
I find it to be a very non-invasive and passive software that works well.

Yes on mono... follow the source.

DM

Archived from groups: rec.audio.pro (More info?)

Bob Cain wrote:
> Looks like a job for ABX!
>
> It's amazing to me that so many people seem to consider it invalid or
> irrelevant. It must be that way or a whole lot of the high end
> arguments that happen within high powered groups could be settled quite
> easily and objectively.

I'm not sure I feel the same way. Certainly ABX has the advantage that
it is unbiased. This is a big advantage. But, that doesn't make it
conclusive. In my living room I have a Panasonic TV that's hooked up
to my stereo. It has an odd noise I can only describe as something
like "aluminum rattling very faintly in the upper midrange"[1], and this
noise only happens when I use the variable-volume audio out (as opposed
to the fixed-volume one). Thing is, I bought this TV while unemployed
and watched DVD movies and other stuff on it quite a lot when it was new,
but it still took me around two weeks to first notice this noise. Now
I can spot it quite easily and in fact have had to learn to tune it out
while watching TV.

So, my question is, if you conducted an ABX test with the two outputs
of my TV, would the test detect a statistically significant difference
between the two? My guess is, unless the test required all the listeners
to listen to listen to many, many hours (like 10 or 20 hours) of audio,
the test wouldn't detect any measurable difference. And yet I'm quite
confident the noise exists.

- Logan