Re: [RDD] Harmonic mixing implementation
I'm involved primarily in Talk programming, so I don't have the level of experience enjoyed by most music programmers. But don't most - if not at least many - contemporary songs these days contain key and tempo information in their tags or chunks? I realize this does not help us on older songs, but it would seem to me that such data would now be a staple in basic commercial music production on its way thru the post-process, for the benefit of radio airplay. -AP ___ Rivendell-dev mailing list Rivendell-dev@lists.rivendellaudio.org http://lists.rivendellaudio.org/mailman/listinfo/rivendell-dev
[RDD] Normalization of audio file
Hi, just a curiosity :-) As many of you I have imported tons of songs in the library normalizing everything at -13dBFS. I read somewhere that this number comes from the conversion between analog (plus headroom) to digital. Is that right? I really trust in you, but don't you think you loose a lot of information by doing this normalization? I know you can vary that value but at the beginning I didn't know I could. But is it a conversion only related to professional audio cards or is it a relation that is valid anywhere? Our pipe is Rivendell - M-Audio Delta 1010 - Soundcraft Mixer.. actually zeros doesn't match (we have to trim a bit) Again.. it is just a curiosity. Thank you Alessio ___ Rivendell-dev mailing list Rivendell-dev@lists.rivendellaudio.org http://lists.rivendellaudio.org/mailman/listinfo/rivendell-dev
Re: [RDD] Normalization of audio file
Nowadays I'd use RMS normalizing or even better R128. Atenciosamente, ** ** *Fernando Della Torre* Tecnologia da Informação (: +55 16 8137-1240 (: +55 16 9137-2886 *: *f...@vdit.com.br* V.D.I.T. Soluções em Virtualização ** ** ** ** A utilização deste e-mail não implica em autorização ou outorga de poderes para seu usuário praticar qualquer ato em nome das empresas citadas, cuja representação considera-se válida se praticada exclusivamente por representante legal ou procurador devidamente constituído, na forma estabelecida em seu respectivo estatuto ou contrato social 2012/12/11 Alessio Elmi alessio_e...@hotmail.com Hi, just a curiosity :-) As many of you I have imported tons of songs in the library normalizing everything at -13dBFS. I read somewhere that this number comes from the conversion between analog (plus headroom) to digital. Is that right? I really trust in you, but don't you think you loose a lot of information by doing this normalization? I know you can vary that value but at the beginning I didn't know I could. But is it a conversion only related to professional audio cards or is it a relation that is valid anywhere? Our pipe is Rivendell - M-Audio Delta 1010 - Soundcraft Mixer.. actually zeros doesn't match (we have to trim a bit) Again.. it is just a curiosity. Thank you Alessio ___ Rivendell-dev mailing list Rivendell-dev@lists.rivendellaudio.org http://lists.rivendellaudio.org/mailman/listinfo/rivendell-dev ___ Rivendell-dev mailing list Rivendell-dev@lists.rivendellaudio.org http://lists.rivendellaudio.org/mailman/listinfo/rivendell-dev
Re: [RDD] Normalization of audio file
On Tuesday 11 December 2012 09:31:05 am Alessio Elmi wrote: Hi, just a curiosity :-) As many of you I have imported tons of songs in the library normalizing everything at -13dBFS. I read somewhere that this number comes from the conversion between analog (plus headroom) to digital. Is that right? Hi there, Alessio OK, here's the deal. ( I am a pro, and this is in my field ) Analog zero, is 1 milliwatt in 600 ohms. It's a telephone standard, and the decibel, 1/10 of a bell, was named for Alexander Graham Bell, the inventor of the telephone. With me so far ? Good. ( yes, later, non-American's decided it would be 1 volt in 10,000 ohms, but it gives you more choices when picking standards, I guess. ) That level was chosen for several reasons. Loud enough for a normal person to hear, not so loud as to cross-talk between wires, a variety of things. Still, this was in the analog days, so going over the limit could cause some problems to begin to appear, but wasn't fatal. Zero VU is a different standard, also based on the human ear. The zero being 1 volt thing takes none of these important factors into consideration. It's just nice, round numbers. Because distortion is important, there is always headroom designed into any gear, because driving a signal against the rail causes clipping, very objectionable distortion. The degree of objection depends on the depth of clip. The peak to average in voice, is about 4:1 on average, and the analog meters show you pretty much the average, so the peak is about 4 times the level the meter shows you, on average. Peaks on a LOUD passage can be much higher. Still, defining a standard level for any operation is important, so that you can achieve some degree of consistency. Consequently, zero db is a decent level, loud enough, yet not too loud, and the nature of analog equipment is somewhat forgiving if you do accidentally occasionally hit the rail. Don't hit it too hard, and the distortion isn't that bad. Now digital. Zero is defined as THE RAIL ! Worse, since it's digitized, zero level is all 1's in the bit stream. There is no headroom, no forgiveness, AT ALL ! Hit zero at any time, for even one peak, and the distortion is near 100% at best. So, although it might appear that -13 leaves a lot on the table, it really doesn't. Even in the analog days, most professional equipment would pass +22 and some +26 before hitting the rail, or it left a good 20 db average head room. Peak headroom was considerably less, because they are peaks. In actual truth, -13 digital is only about half the headroom that the old analog equipment had. Also, db is a relative ratio thing anyway. -13db relative to what ? Well, relative to total, fatal, destruction !! And, that -13 is PEAK level, not average ! More importantly, how many of the available 1's in the sample is that -13 ? It better leave a few, since zero leaves, well, zero !! Does this help ? -- Cowboy http://cowboy.cwf1.com How long a minute is depends on which side of the bathroom door you're on. ___ Rivendell-dev mailing list Rivendell-dev@lists.rivendellaudio.org http://lists.rivendellaudio.org/mailman/listinfo/rivendell-dev
Re: [RDD] Normalization of audio file
Interesting.. @Fernando: so you normalize files before going into Rivendell and disable rd.normaliz? Or you have found a way to combine R128 on-the-fly inside 'rdimport'? @Cowboy: well I was born in the digital era and I am not very confident with analog/electrical measures. I am learning step-by-step... Apart from that do you think that a CDrip wave normalized ad -13dBFS could be compromised? I mean in terms of definitions (quantization of 16 bit depth)... I have an entire catalog that way. However I don't feel the sound being deteriorated, it still sounds clear. Alessio 2012/12/11 Fernando Della Torre f...@vdit.com.br: Nowadays I'd use RMS normalizing or even better R128. Atenciosamente, Fernando Della Torre Tecnologia da Informação (: +55 16 8137-1240 (: +55 16 9137-2886 *: f...@vdit.com.br V.D.I.T. Soluções em Virtualização A utilização deste e-mail não implica em autorização ou outorga de poderes para seu usuário praticar qualquer ato em nome das empresas citadas, cuja representação considera-se válida se praticada exclusivamente por representante legal ou procurador devidamente constituído, na forma estabelecida em seu respectivo estatuto ou contrato social 2012/12/11 Alessio Elmi alessio_e...@hotmail.com Hi, just a curiosity :-) As many of you I have imported tons of songs in the library normalizing everything at -13dBFS. I read somewhere that this number comes from the conversion between analog (plus headroom) to digital. Is that right? I really trust in you, but don't you think you loose a lot of information by doing this normalization? I know you can vary that value but at the beginning I didn't know I could. But is it a conversion only related to professional audio cards or is it a relation that is valid anywhere? Our pipe is Rivendell - M-Audio Delta 1010 - Soundcraft Mixer.. actually zeros doesn't match (we have to trim a bit) Again.. it is just a curiosity. Thank you Alessio ___ Rivendell-dev mailing list Rivendell-dev@lists.rivendellaudio.org http://lists.rivendellaudio.org/mailman/listinfo/rivendell-dev ___ Rivendell-dev mailing list Rivendell-dev@lists.rivendellaudio.org http://lists.rivendellaudio.org/mailman/listinfo/rivendell-dev ___ Rivendell-dev mailing list Rivendell-dev@lists.rivendellaudio.org http://lists.rivendellaudio.org/mailman/listinfo/rivendell-dev
Re: [RDD] Normalization of audio file
On Tue, 2012-12-11 at 12:40 -0500, Cowboy wrote: For me and mine, I'd rather stay away from the certainty of severe distortion running out of head room, in favor of the maybe some ONE out there *might* notice a little noise on the quietest passages. Amen to that. Robert ___ Rivendell-dev mailing list Rivendell-dev@lists.rivendellaudio.org http://lists.rivendellaudio.org/mailman/listinfo/rivendell-dev
Re: [RDD] Normalization of audio file
I reply only to you... When you say anything above -24 is pointless you mean normalizing louder than -24? So even with high dynamic music (like classical for example) you don't feel the risk of music loss? An other interesting thing is that - as I was saying music normalized at -13 (obviously that's the highest peak!) goes out (through Delta 1010) and get the mixer i bit low... VU meter (with no gain) goes around -15/-10... BUT if I normalized a sinus 1000Hz (so constantly -13dBFS) mixer's VU meters go crazy... positive values? Is there anything wrong? Thank you Alessio 2012/12/11 Cowboy c...@cwf1.com: On Tuesday 11 December 2012 11:11:56 am Alessio Elmi wrote: @Cowboy: well I was born in the digital era and I am not very confident with analog/electrical measures. I am learning step-by-step... OK. There is one reason to up the levels. ONE ! That reason is noise. All amplifiers have a noise floor. Pick any audio amp. Disconnect and short out the input. Turn the volume all The WAY UP ! What do you hear ? You hear the self-noise of the amp. THE reason to make the content louder, is to push that noise as far down as possible. The louder the content, the less you turn up the volume, the less noise is *added* by that amp. Apart from that do you think that a CDrip wave normalized ad -13dBFS could be compromised? I mean in terms of definitions (quantization of 16 bit depth)... I have an entire catalog that way. The noise floor limit of 16 bit sampling is -96 dbfs. ( zero = fs ) The lower you normalize, the higher you need your output gain. Normalizing at -13 gives you 83 db dynamic range. ( the usable difference between your peak and the noise floor ) Normalizing higher, and risking SEVERE distortion at FULL volume regardless of where your level is set, gives you more range to the noise floor. So, the question becomes what's the maximum dynamic range of all of the equipment after and including your first DAC ? There is no point whatever in exceeding that, because that noise will be present and above the noise of your playback anyway. Humans can have hearing dynamic range that exceeds 100 db, but that means from can not detect, even in a quiet room, to where hearing stops, and real pain begins. It's theoretically possible under ideal conditions for humans to hear the full limit of 16 bit recording, but that means a quiet room and good ears for the low passages, and pain ( not hearing ) at the loudest passages. Theoretically. In other words, if your feeding a really good broadcast transmitter with a remarkable 72 db dynamic range, 96-72=24 db, so normalizing to anything, repeat ANYTHING above -24 is pointless, because all your doing then is substituting transmitter noise for recorded noise, and reducing head room. You are trading head room, and risk of bad distortion, for blinky lights on your meter. Nothing more, because any gain you have at that stage will be lost in the transmitter noise. In a case like this, that extra 10 db of headroom can save your bacon when a hot jock drives things to the wall. If you're broadcasting over internet, then your maximum range is the typical sound card in consumer gear. About 35 db, give or take. For me and mine, I'd rather stay away from the certainty of severe distortion running out of head room, in favor of the maybe some ONE out there *might* notice a little noise on the quietest passages. -- Cowboy http://cowboy.cwf1.com How long a minute is depends on which side of the bathroom door you're on. ___ Rivendell-dev mailing list Rivendell-dev@lists.rivendellaudio.org http://lists.rivendellaudio.org/mailman/listinfo/rivendell-dev ___ Rivendell-dev mailing list Rivendell-dev@lists.rivendellaudio.org http://lists.rivendellaudio.org/mailman/listinfo/rivendell-dev
[RDD] The Great Audio Voltage Debate quick ref chart
Noticed lengthy discussions on normalization and found this chart that might be of interest http://www.google.ca/url?sa=trct=jq=0%20dbmsource=webcd=4sqi=2ved=0CD4QFjADurl=http%3A%2F%2Fwww.tecmag.com%2Fpdf%2Fdbm_v.pdfei=fN_HUK2TGIaE2gXlxoCgAQusg=AFQjCNEY25QxasGndqL5ZAXkYDCgFaB4KQcad=rja also this explanation I found useful http://en.wikipedia.org/wiki/DBm Audacity shows a scale with 1.0 as a max usually but that is PEAK only and is actually 2.0 VAC Peak to Peak RMS is 0.775 and average is closer to 0.5 so be sure to compare apples to apples and oranges to oranges I have a portable oscillator for aligning my mixer board and all other audio sections including the PC's but only have a standard digital multimeter and or a standard analog mutimeter neither of which are calibrated to read in db, dbv or dbm unlike a lot of pro audio test sets are so these relationships help to be able to use the basic meters to calibrate my systems; but one has to be very aware of the relationships and not to confuse, peak, peak-to-peak, average and RMS values in order to maintain a maximum peak range between -8 dbm and -12 dbm and still retain a further 4 db of additional headroom for protection. ___ Rivendell-dev mailing list Rivendell-dev@lists.rivendellaudio.org http://lists.rivendellaudio.org/mailman/listinfo/rivendell-dev
Re: [RDD] The Great Audio Voltage Debate quick ref chart
1.0 in Audacity is 0 dBFS. This has nothing to do with voltage. It has only a tenuous relationship to 0 dBVU. Various people use various values for 0 dBVU. Usually it's -18 (EBU recommendation) or -20 (AES recommendation) down from 0 dBFS. Now we kinda get to voltages. 0dBVU is +4 dBM (1.23V RMS at 600 ohms) in pro systems. 0 dBVU is -10 dBV (.316V RMS). Bill On Dec 11, 2012, at 17:47, VE4PER/ Andy ve4...@aim.com wrote: Noticed lengthy discussions on normalization and found this chart that might be of interest http://www.google.ca/url?sa=trct=jq=0%20dbmsource=webcd=4sqi=2ved=0CD4QFjADurl=http%3A%2F%2Fwww.tecmag.com%2Fpdf%2Fdbm_v.pdfei=fN_HUK2TGIaE2gXlxoCgAQusg=AFQjCNEY25QxasGndqL5ZAXkYDCgFaB4KQcad=rja also this explanation I found useful http://en.wikipedia.org/wiki/DBm Audacity shows a scale with 1.0 as a max usually but that is PEAK only and is actually 2.0 VAC Peak to Peak RMS is 0.775 and average is closer to 0.5 so be sure to compare apples to apples and oranges to oranges I have a portable oscillator for aligning my mixer board and all other audio sections including the PC's but only have a standard digital multimeter and or a standard analog mutimeter neither of which are calibrated to read in db, dbv or dbm unlike a lot of pro audio test sets are so these relationships help to be able to use the basic meters to calibrate my systems; but one has to be very aware of the relationships and not to confuse, peak, peak-to-peak, average and RMS values in order to maintain a maximum peak range between -8 dbm and -12 dbm and still retain a further 4 db of additional headroom for protection. ___ Rivendell-dev mailing list Rivendell-dev@lists.rivendellaudio.org http://lists.rivendellaudio.org/mailman/listinfo/rivendell-dev ___ Rivendell-dev mailing list Rivendell-dev@lists.rivendellaudio.org http://lists.rivendellaudio.org/mailman/listinfo/rivendell-dev
Re: [RDD] Normalization of audio file
On Tue, 2012-12-11 at 15:31 +0100, Alessio Elmi wrote: but don't you think you loose a lot of information by doing this normalization? Alessio, setting up the programme chain in a radio or tv or anything else is a constant trade off between the highest electrical level the sound will produce, and the ability of the system to reproduce this faithfully. Cowboy is right. In the Analogue days A amps and B Amps were set up to handle at least 20db of overload before distorting. We monitored using 'VU Meters' which by their very design showed an average of the signal, and definitely did not show peak audio levels. There were 'peak programme meters' which were faster, harder to follow, and they too did not show all the peaks. The only really good meters we had were oscilloscopes which could show voltage peaks and I can confirm that '0 on the meter' included a whole lot of data that went well beyond that. All the amplifiers in the system were designed to have as much headroom as possible and even when you did overload them the onset of distortion was gentle, so those peaks were not so noticeable if they did distort. The combination of vacuum tubes and the circuitry surrounding them that allowed for variations in performance as the tube aged meant the distortion crept in as you reached overload. That is not so in the 'digital' era where there is *NO* margin above whatever headroom you have. Trading 13 db of signal to noise to avoid distortion is not a bad trade off because audio peaking too high goes from no distortion to pretty much 100% distortion instantly. at the noise end of the dynamic range 13db is not a big deal because most of what we get from CD is well within 60db range and the CD standard has a noise floor below 73db so we're still 'in with a grin' by about 20db The dynamic range of 16 bit 44100 audio is far greater then the dynamic range of anything in the transmission path, so there will be some gain manipulation in the path. By setting -13 we have a system which gives a reasonable margin before distortion. OK in a controlled studio situation. Recording live with digital gear I tend to run closer to -20 to allow for the unexpected peaks that just come along. In a perfect world you could run all your audio at 0, but having looked at a few spectral analysis read outs I can tell you those pesky peaks just happen along and ruin your day. Once you have distortion you can't get rid of it. I think you are confusing 'normalisation' with what is called 'compression' [mp3 etc] but is in fact selecting just enough to retain a semblance of the original but in a smaller file which in my view is an electronic sham. There are places where 'compressed' audio is fine. We use mp3 for news tracks. Reducing file size for distribution make sense. In the compromise that all audio broadcasting involves, setting the normalisation level at -13 is not a bad decision. The sound card will turn this into anything from -20dBm to +4dBm which then goes to the desk, if you run that way, or may be routed as digital to a digital mixer. Yes the output will be 13 db below the maximum the sound card can deliver, but it won't be distorted. regards Robert Jeffares Big Valley Radio Thames ___ Rivendell-dev mailing list Rivendell-dev@lists.rivendellaudio.org http://lists.rivendellaudio.org/mailman/listinfo/rivendell-dev
Re: [RDD] The Great Audio Voltage Debate quick ref chart
http://www.sengpielaudio.com/calculator-db-volt.htm On 12/12/12 04:15, Bill Putney wrote: 1.0 in Audacity is 0 dBFS. This has nothing to do with voltage. It has only a tenuous relationship to 0 dBVU. Various people use various values for 0 dBVU. Usually it's -18 (EBU recommendation) or -20 (AES recommendation) down from 0 dBFS. Now we kinda get to voltages. 0dBVU is +4 dBM (1.23V RMS at 600 ohms) in pro systems. 0 dBVU is -10 dBV (.316V RMS). Bill On Dec 11, 2012, at 17:47, VE4PER/ Andy ve4...@aim.com wrote: Noticed lengthy discussions on normalization and found this chart that might be of interest http://www.google.ca/url?sa=trct=jq=0%20dbmsource=webcd=4sqi=2ved=0CD4QFjADurl=http%3A%2F%2Fwww.tecmag.com%2Fpdf%2Fdbm_v.pdfei=fN_HUK2TGIaE2gXlxoCgAQusg=AFQjCNEY25QxasGndqL5ZAXkYDCgFaB4KQcad=rja also this explanation I found useful http://en.wikipedia.org/wiki/DBm Audacity shows a scale with 1.0 as a max usually but that is PEAK only and is actually 2.0 VAC Peak to Peak RMS is 0.775 and average is closer to 0.5 so be sure to compare apples to apples and oranges to oranges I have a portable oscillator for aligning my mixer board and all other audio sections including the PC's but only have a standard digital multimeter and or a standard analog mutimeter neither of which are calibrated to read in db, dbv or dbm unlike a lot of pro audio test sets are so these relationships help to be able to use the basic meters to calibrate my systems; but one has to be very aware of the relationships and not to confuse, peak, peak-to-peak, average and RMS values in order to maintain a maximum peak range between -8 dbm and -12 dbm and still retain a further 4 db of additional headroom for protection. ___ Rivendell-dev mailing list Rivendell-dev@lists.rivendellaudio.org http://lists.rivendellaudio.org/mailman/listinfo/rivendell-dev ___ Rivendell-dev mailing list Rivendell-dev@lists.rivendellaudio.org http://lists.rivendellaudio.org/mailman/listinfo/rivendell-dev ___ Rivendell-dev mailing list Rivendell-dev@lists.rivendellaudio.org http://lists.rivendellaudio.org/mailman/listinfo/rivendell-dev
Re: [RDD] Normalization of audio file
On Tue, 2012-12-11 at 20:06 +0100, Alessio Elmi wrote: An other interesting thing is that - as I was saying music normalized at -13 (obviously that's the highest peak!) goes out (through Delta 1010) and get the mixer i bit low... VU meter (with no gain) goes around -15/-10... BUT if I normalized a sinus 1000Hz (so constantly -13dBFS) mixer's VU meters go crazy... positive values? Is there anything wrong? very definitely. If the tone in the cart is -13 the output should be stable and you should peak this at '0' then compare music level set to same level. Robert ___ Rivendell-dev mailing list Rivendell-dev@lists.rivendellaudio.org http://lists.rivendellaudio.org/mailman/listinfo/rivendell-dev