Mu-law decreases noise at low levels but makes the audio vulnerable to clipping. Is there any codec that does the opposite? I am looking for one that gives rich quality at loud levels but has the drawback of a poor s/n ratio. Any hope?
Opposite of Mu-law?
Started by ●May 18, 2004
Reply by ●May 18, 20042004-05-18
"Curious" <curious11112001@yahoo.com> wrote in message news:34a4f456.0405171918.8263e25@posting.google.com...> Mu-law decreases noise at low levels but makes the audio vulnerable to > clipping. Is there any codec that does the opposite? > > I am looking for one that gives rich quality at loud levels but has > the drawback of a poor s/n ratio. Any hope?I'd go with linear PCM with enough dither to mask the correlated noise. Even at four bits, dither helps. Dynamic range would be about 24 dB (48 dB for 8 bits), but it would be "clean" except for the noise. If you didn't use dither, the sound would be distorted and grainy.
Reply by ●May 18, 20042004-05-18
"Karl Uppiano" <karl_uppiano@verizon.net> wrote in message news:<YDfqc.81567$sK3.30189@nwrddc03.gnilink.net>...> "Curious" <curious11112001@yahoo.com> wrote in message > news:34a4f456.0405171918.8263e25@posting.google.com... > > Mu-law decreases noise at low levels but makes the audio vulnerable to > > clipping. Is there any codec that does the opposite? > > > > I am looking for one that gives rich quality at loud levels but has > > the drawback of a poor s/n ratio. Any hope? > > I'd go with linear PCM with enough dither to mask the correlated noise. Even > at four bits, dither helps. Dynamic range would be about 24 dB (48 dB for 8 > bits), but it would be "clean" except for the noise. If you didn't use > dither, the sound would be distorted and grainy.Adobe Auditon does allow converting to linear PCM format. Dithering is also possible w/ AA. If I want more of this "anti-Mu Law" effect should I dither more bits?
Reply by ●May 18, 20042004-05-18
On 17 May 2004 20:18:22 -0700, curious11112001@yahoo.com (Curious) wrote:>Mu-law decreases noise at low levels but makes the audio vulnerable to >clipping. Is there any codec that does the opposite? > >I am looking for one that gives rich quality at loud levels but has >the drawback of a poor s/n ratio. Any hope?Not entirely sure what you mean about clipping - Mu law is just a level weighting within the operating range of the DAC. Once you hit the biggest number, you clip - whatever the law. d Pearce Consulting http://www.pearce.uk.com
Reply by ●May 18, 20042004-05-18
Curious wrote:> "Karl Uppiano" <karl_uppiano@verizon.net> wrote in message news:<YDfqc.81567$sK3.30189@nwrddc03.gnilink.net>... > >>"Curious" <curious11112001@yahoo.com> wrote in message >>news:34a4f456.0405171918.8263e25@posting.google.com... >> >>>Mu-law decreases noise at low levels but makes the audio vulnerable to >>>clipping. Is there any codec that does the opposite? >>> >>>I am looking for one that gives rich quality at loud levels but has >>>the drawback of a poor s/n ratio. Any hope? >> >>I'd go with linear PCM with enough dither to mask the correlated noise. Even >>at four bits, dither helps. Dynamic range would be about 24 dB (48 dB for 8 >>bits), but it would be "clean" except for the noise. If you didn't use >>dither, the sound would be distorted and grainy. > > > Adobe Auditon does allow converting to linear PCM format. Dithering is > also possible w/ AA. > > If I want more of this "anti-Mu Law" effect should I dither more bits?Mu-law is a compression standard. The transfer function is IIRC an approximate hyperbolic cosine. It is used to compress a signal for transmission. Its inverse (again IIRC), an arc hyperbolic cosine, is used at the receiving end to restore (expand) the original signal. Note that "compress" and "expand" don't refer here to the usual audio compander pair which works on volume, but to instantaneous values of the signal. What you want to do sounds interesting, but either I don't fully understand you, or you don't understand what mu-law is all about. Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
Reply by ●May 18, 20042004-05-18
Opposite of mu law is mu law isn't it? Instead of compressing , transmitting then expanding you expand, transmit and then compress. Best of Luck - Mike "Don Pearce" <donald@pearce.uk.com> wrote in message news:40aa383a.5429015@news.plus.net...> On 17 May 2004 20:18:22 -0700, curious11112001@yahoo.com (Curious) > wrote: > > >Mu-law decreases noise at low levels but makes the audio vulnerable to > >clipping. Is there any codec that does the opposite? > > > >I am looking for one that gives rich quality at loud levels but has > >the drawback of a poor s/n ratio. Any hope? > > Not entirely sure what you mean about clipping - Mu law is just a > level weighting within the operating range of the DAC. Once you hit > the biggest number, you clip - whatever the law. > > d > Pearce Consulting > http://www.pearce.uk.com
Reply by ●May 18, 20042004-05-18
Got a "cannot find server" error so posting again. Sorry for resulting multi-posts. Jerry Avins <jya@ieee.org> wrote in message news:<40aa5787$0$3013$61fed72c@news.rcn.com>...> Curious wrote: > > > "Karl Uppiano" <karl_uppiano@verizon.net> wrote in message news:<YDfqc.81567$sK3.30189@nwrddc03.gnilink.net>... > > > >>"Curious" <curious11112001@yahoo.com> wrote in message > >>news:34a4f456.0405171918.8263e25@posting.google.com... > >> > >>>Mu-law decreases noise at low levels but makes the audio vulnerable to > >>>clipping. Is there any codec that does the opposite? > >>> > >>>I am looking for one that gives rich quality at loud levels but has > >>>the drawback of a poor s/n ratio. Any hope? > >> > >>I'd go with linear PCM with enough dither to mask the correlated noise. Even > >>at four bits, dither helps. Dynamic range would be about 24 dB (48 dB for 8 > >>bits), but it would be "clean" except for the noise. If you didn't use > >>dither, the sound would be distorted and grainy. > > > > > > Adobe Auditon does allow converting to linear PCM format. Dithering is > > also possible w/ AA. > > > > If I want more of this "anti-Mu Law" effect should I dither more bits? > > Mu-law is a compression standard. The transfer function is IIRC an > approximate hyperbolic cosine. It is used to compress a signal for > transmission. Its inverse (again IIRC), an arc hyperbolic cosine, is > used at the receiving end to restore (expand) the original signal. > > Note that "compress" and "expand" don't refer here to the usual audio > compander pair which works on volume, but to instantaneous values > of the signal. > > What you want to do sounds interesting, but either I don't fully > understand you, or you don't understand what mu-law is all about. > > JerryWhat I would like is to increase the clipping point - that is, the increase the loudnest sound that can be recorded without causing distortion. I would like a type of codec that does this. Has a strength of being able to handle louder sounder w/out clipping, yet with the price of decreased SNR. Mu-law seems to do the opposite. BTW I tried dithering with AA, it did *not* increase the clipping point at all. The dithering *does* makes the sound less "choppy" at lower levels. This is not what I was looking for though. I am looking for something that increases the ability to handle louder sounds.
Reply by ●May 19, 20042004-05-19
Curious wrote:> Got a "cannot find server" error so posting again. Sorry for resulting > multi-posts. > > Jerry Avins <jya@ieee.org> wrote in message news:<40aa5787$0$3013$61fed72c@news.rcn.com>......>>Mu-law is a compression standard. The transfer function is IIRC an >>approximate hyperbolic cosine. It is used to compress a signal for >>transmission. Its inverse (again IIRC), an arc hyperbolic cosine, is >>used at the receiving end to restore (expand) the original signal. >> >>Note that "compress" and "expand" don't refer here to the usual audio >>compander pair which works on volume, but to instantaneous values >>of the signal. >> >>What you want to do sounds interesting, but either I don't fully >>understand you, or you don't understand what mu-law is all about. >> >>Jerry > > > What I would like is to increase the clipping point - that is, the > increase the loudnest sound that can be recorded without causing > distortion.There are two ways to do that: increase the number of bits, or use something like mu-law. Mu-law allows a channel with only eight bits to pass 12-bit levels without clipping.> I would like a type of codec that does this. Has a > strength of being able to handle louder sounder w/out clipping, yet > with the price of decreased SNR. Mu-law seems to do the opposite.No. That's exactly what mu-law was designed to do. The price is not only SNR, but a modest amount of non-linear distortion. With voice signals -- mu-law's intended material -- primarily intermodulation distortion. You could use a more severe compression curve that allows, say 14 bits before clipping, but the distortion will be higher.> BTW I tried dithering with AA, it did *not* increase the clipping > point at all. The dithering *does* makes the sound less "choppy" at > lower levels. This is not what I was looking for though.Why did you think that dithering would allow higher levels? There's profit in answering this. Think about how dithering affects dynamic range.> I am looking for something that increases the ability to handle louder > sounds.Halve the gain so loud sounds don't overload. Use dithering to gain the bit back as the low end. Then mu-compress the dithered signal. Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
Reply by ●May 19, 20042004-05-19
"Curious" <curious11112001@yahoo.com> wrote in message news:34a4f456.0405181607.2613281b@posting.google.com...> > I am looking for something that increases the ability to handle louder > sounds.Just do like the guy in "Spinal Tap" and get an amp that goes to eleven ;-) Clay. p.s. Actually mu-law gives about 35 dB of S/N of about a 60 dB range. Not bad for just 8 bits. You can extend this idea to whatever you need.
Reply by ●May 19, 20042004-05-19
"Curious" <curious11112001@yahoo.com> wrote in message news:34a4f456.0405180814.271ffc3f@posting.google.com...> "Karl Uppiano" <karl_uppiano@verizon.net> wrote in messagenews:<YDfqc.81567$sK3.30189@nwrddc03.gnilink.net>...> > "Curious" <curious11112001@yahoo.com> wrote in message > > news:34a4f456.0405171918.8263e25@posting.google.com... > > > Mu-law decreases noise at low levels but makes the audio vulnerable to > > > clipping. Is there any codec that does the opposite? > > > > > > I am looking for one that gives rich quality at loud levels but has > > > the drawback of a poor s/n ratio. Any hope? > > > > I'd go with linear PCM with enough dither to mask the correlated noise.Even> > at four bits, dither helps. Dynamic range would be about 24 dB (48 dBfor 8> > bits), but it would be "clean" except for the noise. If you didn't use > > dither, the sound would be distorted and grainy. > > Adobe Auditon does allow converting to linear PCM format. Dithering is > also possible w/ AA. > > If I want more of this "anti-Mu Law" effect should I dither more bits?No, for linear encoding, triangular probability density dither at the 1-bit level is optimal. Adding more dither than that simply adds excessive noise. Adding less dither creates less noise, but leaves more distortion. The clipping level stays the same regardless of the dither level. In fact, if linear *or* Mu-law encoders and decoders are calibrated so that digital full scale has the same analog reference level, the clipping levels will be identical, so the clipping question becomes moot. No matter what, you won't get something for nothing. Mu-law is just a way to re-allocate bits so that they're closer together where the signal is smaller, and farther apart where the signal is larger. It tends to improve the perceived audio quality, but it's hardly what you'd call high-fidelity in its typical implementation. Mu-law was developed so that 8-bit telecommunication channels could transmit voice messages with acceptable intelligibility with minimal processing. It is a very primitive form of audio data compression. There are much better ways of reducing bandwidth these days -- MP3 for example. Even the lowest bit-rate MP3 encoding should sound better than any of the low bit-rate formats discussed previously, assuming you have the necessary codecs and processing horsepower available. What is your application and/or goal? If you're looking for highest quality and lowest bit-rate, you should consider recording with at least 16-bit linear PCM at 44.1 KHz, and then convert it to MP3 or WMA at the highest bit-rate you can afford. On the other hand, if you're looking for an "effect" of some kind, I'd need a lot more information.
