[Jin | Cinemasports]

I need to convert music from 16 bit to 8 bit in order to play it over
a phone system. Does anybody have any practical production advice to
keep the quality as high as possible? When I just reduce the bit
depth, I get a lot of hiss and distortion, so I'm looking for
techniques such as compression or eq to keep it sounding as good as
possible. For example, in audacity what filters should I use before I
reduce the bit depth?

I'm looking for practical production recipes that help maintain audio
quality.

[[email protected]]

On Jan 15, 2:12*pm, "Jin | Cinemasports"
wrote:
I need to convert music from 16 bit to 8 bit in order to play it over
a phone system.

Why?

Does anybody have any practical production advice to
keep the quality as high as possible?

Yes, don't do it. Keep the bit depth at 16 bits. And don't buy into
300 to 3400 Hz for band limiting either.

When I just reduce the bit
depth, *I get a lot of hiss and distortion, so I'm looking for
techniques such as compression or eq to keep it sounding as good as
possible. For example, in audacity what filters should I use before I
reduce the bit depth?

It's possible to minimize the damage but be prepared for a lot, and I
do mean A LOT, of hand processing. 8 bits is a real pain to work with.
I know. I've made tens of thousands of voice prompts for 8 bit
playback in 30+ languages. The best results are obtained by
experienced hand processing.

I'm looking for practical production recipes that help maintain audio
quality.

There is no simple recipe. It depends on the spectral distribution of
the voice recorded. It depends on the intended playback system
characteristics. It depends on the quality of the original recording.
I keep a rack full of compressors, equalizers, DeEssers, Limiters both
digital and analog, and a pile of plug-ins of the same including some
magic juju boxes, much as a chef might keep a rack full of spices.
After evaluating the source recording and playback system
characteristics I choose which to use based on a lot of experience.
I've had folks tell me my results sound almost as good as 16 bits but
in my opinion when it is all said and done 16 bits still beats the
pants off the magic faery dusted 8 bit results.

bobs

Bob Smith
BS Studios / SoundSmith Labs
we organize chaos
http://www.bsstudios.com

[Scott Dorsey]

Jin | Cinemasports wrote:
I need to convert music from 16 bit to 8 bit in order to play it over
a phone system. Does anybody have any practical production advice to
keep the quality as high as possible? When I just reduce the bit
depth, I get a lot of hiss and distortion, so I'm looking for
techniques such as compression or eq to keep it sounding as good as
possible. For example, in audacity what filters should I use before I
reduce the bit depth?

I'm looking for practical production recipes that help maintain audio
quality.

Which encoding are you using? 8 bit uLaw should not have lots of hiss
and distortion although it's not the most wonderful thing around.

Do your processing while monitoring to the uLaw encoded signal, while
listening on a telephone. Tweak it to sound good through that chain
and you're good to go. The key is to monitor the encoded signal.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[Laurence Payne]

On Tue, 15 Jan 2008 16:12:33 -0800 (PST), wrote:

I need to convert music from 16 bit to 8 bit in order to play it over
a phone system.

Why?

I expect he's got one of those auto-answer systems that's fussy about
the sound files it will accept. What do you expect - full
information in a Usenet query? :-)

[Peter Larsen[_2_]]

Laurence Payne wrote:

On Tue, 15 Jan 2008 16:12:33 -0800 (PST), wrote:

I need to convert music from 16 bit to 8 bit in order to play it
over a phone system.

Why?

I expect he's got one of those auto-answer systems that's fussy about
the sound files it will accept. What do you expect - full
information in a Usenet query? :-)

Laurence, that is not how we do it. First we follow up about all possible 8
bit conversions, next someone asks whether he has a license to distribute
said music like that, and eventually we will be told that it is for his
private answering machine, in which case holding a headphone close to its
mic and pressing record gives a passably good result.

Kind regards

Peter Larsen

[Don Pearce]

On Tue, 15 Jan 2008 14:12:58 -0800 (PST), "Jin | Cinemasports"
wrote:

I need to convert music from 16 bit to 8 bit in order to play it over
a phone system. Does anybody have any practical production advice to
keep the quality as high as possible? When I just reduce the bit
depth, I get a lot of hiss and distortion, so I'm looking for
techniques such as compression or eq to keep it sounding as good as
possible. For example, in audacity what filters should I use before I
reduce the bit depth?

I'm looking for practical production recipes that help maintain audio
quality.

For phones you must use a mu-law or a-law conversion (depending on the
standard in your country). Provided you use dither, that will minimize
both distortion and noise. You may want to compress the dynamics quite
a lot before conversion too.

d

--
Pearce Consulting
http://www.pearce.uk.com

[[email protected]]

On Jan 15, 4:12*pm, wrote:
On Jan 15, 2:12*pm, "Jin | Cinemasports"
wrote:

I need to convert music from 16 bit to 8 bit in order to play it over
a phone system.

Why?


I've had folks tell me my results sound almost as good as 16 bits but
in my opinion when it is all said and done 16 bits still beats the
pants off the magic faery dusted 8 bit results.

bobs

Please note that all my comments were directed at 16 bit pcm vs. 8 bit
pcm the latter of which should be outlawed in this day of inexepensive
16 ~ 24 pcm D/As. uLaw or ALaw encoding is a different animal
altogether. As noted, it would help to know the tech details of the
OP's target playback system.

bobs

Bob Smith
BS Studios / SoundSmith Labs
we organize chaos
http://www.bsstudios.com

[Tobiah]

For phones you must use a mu-law or a-law conversion (depending on the
standard in your country). Provided you use dither, that will minimize
both distortion and noise. You may want to compress the dynamics quite
a lot before conversion too.

d


Those basically already are compression, or so it seems:

http://en.wikipedia.org/wiki/Ulaw

--
Posted via a free Usenet account from http://www.teranews.com

[Don Pearce]

On Wed, 16 Jan 2008 11:42:55 -0800, Tobiah wrote:



For phones you must use a mu-law or a-law conversion (depending on the
standard in your country). Provided you use dither, that will minimize
both distortion and noise. You may want to compress the dynamics quite
a lot before conversion too.

d


Those basically already are compression, or so it seems:

http://en.wikipedia.org/wiki/Ulaw

No, they aren't compression. Maybe you could call them compansion,
since the idea is that what you do at the sending end gets unwound at
the receiving end.

d

--
Pearce Consulting
http://www.pearce.uk.com

[Ron Capik]

Don Pearce wrote:

On Wed, 16 Jan 2008 11:42:55 -0800, Tobiah wrote:



For phones you must use a mu-law or a-law conversion (depending on the
standard in your country). Provided you use dither, that will minimize
both distortion and noise. You may want to compress the dynamics quite
a lot before conversion too.

d


Those basically already are compression, or so it seems:

http://en.wikipedia.org/wiki/Ulaw

No, they aren't compression. Maybe you could call them compansion,
since the idea is that what you do at the sending end gets unwound at
the receiving end.

d

--
Pearce Consulting
http://www.pearce.uk.com

I'd tend to disagree, in that all compression techniques
reduce the data rate and get "unwound" at the other end.
Thus, since said method packs more [useful] information
into fewer bits, I would tend to call it a compression technique.

If you disagree, then what is your working definition of [data]
compression?


Later...

Ron Capik
--

[Peter Larsen[_2_]]

Ron Capik wrote:

Don Pearce wrote:

On Wed, 16 Jan 2008 11:42:55 -0800, Tobiah wrote:

For phones you must use a mu-law or a-law conversion (depending on
the standard in your country). Provided you use dither, that will
minimize both distortion and noise. You may want to compress the
dynamics quite a lot before conversion too.

d

Those basically already are compression, or so it seems:

http://en.wikipedia.org/wiki/Ulaw

No, they aren't compression. Maybe you could call them compansion,
since the idea is that what you do at the sending end gets unwound at
the receiving end.

d

I'd tend to disagree, in that all compression techniques
reduce the data rate and get "unwound" at the other end.
Thus, since said method packs more [useful] information
into fewer bits, I would tend to call it a compression technique.

If you disagree, then what is your working definition of [data]
compression?

The term to use is bit rate reduction, otherwise we get too much confusion
and misreading.

Ron Capik


Kind regards

Peter Larsen

[Tobiah[_2_]]

Ron Capik wrote:
Don Pearce wrote:

On Wed, 16 Jan 2008 11:42:55 -0800, Tobiah wrote:

For phones you must use a mu-law or a-law conversion (depending on the
standard in your country). Provided you use dither, that will minimize
both distortion and noise. You may want to compress the dynamics quite
a lot before conversion too.

d

Those basically already are compression, or so it seems:

http://en.wikipedia.org/wiki/Ulaw
No, they aren't compression. Maybe you could call them compansion,
since the idea is that what you do at the sending end gets unwound at
the receiving end.

d

--
Pearce Consulting
http://www.pearce.uk.com

I'd tend to disagree, in that all compression techniques
reduce the data rate and get "unwound" at the other end.
Thus, since said method packs more [useful] information
into fewer bits, I would tend to call it a compression technique.

If you disagree, then what is your working definition of [data]
compression?


I was thinking of compression in the sense of making louder parts
softer and making softer parts louder. The curves of
the algorithm look like a base amplification with a soft knee
compression.

I was not considering however that this all gets undone upon playing
the sound. The whole thing is not really compression, but more of
whatever class Dolby is in, say.



--
Posted via a free Usenet account from http://www.teranews.com

[Ron Capik]

Peter Larsen wrote:

Ron Capik wrote:

Don Pearce wrote:

On Wed, 16 Jan 2008 11:42:55 -0800, Tobiah wrote:

For phones you must use a mu-law or a-law conversion (depending on
the standard in your country). Provided you use dither, that will
minimize both distortion and noise. You may want to compress the
dynamics quite a lot before conversion too.

d

Those basically already are compression, or so it seems:

http://en.wikipedia.org/wiki/Ulaw

No, they aren't compression. Maybe you could call them compansion,
since the idea is that what you do at the sending end gets unwound at
the receiving end.

d

I'd tend to disagree, in that all compression techniques
reduce the data rate and get "unwound" at the other end.
Thus, since said method packs more [useful] information
into fewer bits, I would tend to call it a compression technique.

If you disagree, then what is your working definition of [data]
compression?

The term to use is bit rate reduction, otherwise we get too much confusion
and misreading.

Ron Capik

Kind regards

Peter Larsen

If the data base is reduced the needed bit rate would also reduced. Would
like to add latency to the definition?


Later...

Ron Capik
--

[Don Pearce]

On Thu, 17 Jan 2008 00:49:17 GMT, Ron Capik
wrote:

Don Pearce wrote:

On Wed, 16 Jan 2008 11:42:55 -0800, Tobiah wrote:



For phones you must use a mu-law or a-law conversion (depending on the
standard in your country). Provided you use dither, that will minimize
both distortion and noise. You may want to compress the dynamics quite
a lot before conversion too.

d


Those basically already are compression, or so it seems:

http://en.wikipedia.org/wiki/Ulaw

No, they aren't compression. Maybe you could call them compansion,
since the idea is that what you do at the sending end gets unwound at
the receiving end.

d

--
Pearce Consulting
http://www.pearce.uk.com

I'd tend to disagree, in that all compression techniques
reduce the data rate and get "unwound" at the other end.
Thus, since said method packs more [useful] information
into fewer bits, I would tend to call it a compression technique.

If you disagree, then what is your working definition of [data]
compression?


OK, we're talking different compressions here. In my original reply,
when I said he might want to compress it first, I meant the dynamics,
so the 8 bits wouldn't be called on to handle too much dynamic range.

Given that you have only 8 bits, and they aren't really sufficient for
high quality transmission, you have to make the best of them and that
is where mu and A law come in. By making the least significant bits
smaller steps than the most significant ones, you lower the noise
floor and win lower quantization distortion at low levels at the
expense of a slight increase in QD all the way up. The tradeoff is
worth it. When I said that what is done at the beginning is unwound at
the end, I meant that the receiving codec is a mirror image of the
transmitting one, so the net result is no compression.

And finally, decimating to 8 bits is not even data compression, just
data reduction, which is quite different.

d
--
Pearce Consulting
http://www.pearce.uk.com