-
Get involved.
We want your input!
Apply for Membership and join the conversations about everything related to broadcasting.
After we receive your registration, a moderator will review it. After your registration is approved, you will be permitted to post.
If you use a disposable or false email address, your registration will be rejected.
After your membership is approved, please take a minute to tell us a little bit about yourself.
https://www.radiodiscussions.com/forums/introduce-yourself.1088/
Thanks in advance and have fun!
RadioDiscussions Administrators
Digital Modulation Question
- Thread starter Len14043
- Start date
BOTTOM LINE: The digital medium does not permit excess amplitude or “over-modulation”. Theoretical “digital zero-VU” is an ABSOLUTE LIMIT... There is NO “graceful” transition into escalating distortion [as is the case with a saturated analog tape recording or an amplifier driven toward the point of waveform clipping]. Even the very-slightest incursion into the “red-zone” will result in MASSIVE distortion due to a literal bit-conversion shutdown. It’s also interesting to note that [shy of that maximum point], digital distortion is actually HIGHER at levels well-below maximum – the OPPOSITE of the condition experienced in the analog domain.
What you are noticing is the result of prudent and effective AUDIO PROCESSING [GASP :
]—FINALLY, Thank God! My overall impression of nearly-every HD-2 service I have heard is similar to that of an unsophisticated one-stoplight radio station operated by Ma & Pa Kettle in the sticks. They are wimpy and lack-luster. Digital audio BENEFITS from HIGH LEVELS of JUDICIOUS processing [that statement is bound to invoke a shooting match here]. This is evidenced in most startling detail by the improvement in latter-issue CD re-masters. They sound better than the earlier digital “first tries” due to advanced signal control which allows the audio to “hug” the maximum level point.
This becomes increasingly-pronounced when a lower bitrate CODEC is involved [HD-2 services are at 48kbps]. Contrary to another contention here that bit-reduced digital audio streams sound better when they are not heavily-processed; HD-2’s formerly-un-compelling audio signature is likely the result of that misplaced philosophy. The best sounding sub-48k streams I have originated and/or heard are ones that benefit from agile volume compression, limiting, and the proverbial heavy finger on the “more” button!
What you are noticing is the result of prudent and effective AUDIO PROCESSING [GASP :
]—FINALLY, Thank God! My overall impression of nearly-every HD-2 service I have heard is similar to that of an unsophisticated one-stoplight radio station operated by Ma & Pa Kettle in the sticks. They are wimpy and lack-luster. Digital audio BENEFITS from HIGH LEVELS of JUDICIOUS processing [that statement is bound to invoke a shooting match here]. This is evidenced in most startling detail by the improvement in latter-issue CD re-masters. They sound better than the earlier digital “first tries” due to advanced signal control which allows the audio to “hug” the maximum level point.This becomes increasingly-pronounced when a lower bitrate CODEC is involved [HD-2 services are at 48kbps]. Contrary to another contention here that bit-reduced digital audio streams sound better when they are not heavily-processed; HD-2’s formerly-un-compelling audio signature is likely the result of that misplaced philosophy. The best sounding sub-48k streams I have originated and/or heard are ones that benefit from agile volume compression, limiting, and the proverbial heavy finger on the “more” button!
M
Mike Walker
Guest
Of course you're going to get some opposition to that, Hipporadio. "Hugging the absolute maximum point", means you start out LOUD! Music is about dynamic contrasts. If you start out as loud as it gets, there's no way to build to a crescendo. Sudden drum THWACKS lose impact, because they're no louder than the rest of the content. There's no "jump" factor, because the only way to "hug the absolute maximum point" is to clip those initial transients that make real music sound REAL!
16 bit digital audio is capable of a maximum dynamic range of 96db (6db pre bit x 16). More than that, actually, as "dither" (low level noise deliberately added to the signal) keeps the "least significant bit" from ever toggling off...which actually makes it possible to reproduce sound BELOW the theoretical -96db point. NO, I'm not suggesting that anybody encode audio that low! Nor am I a purist who believes in no processing. But stations must fine a judicious compromise between having some real dynamics, and consistent average level. Think heavy-handed processing gives more "impact"? Listen to the instruments that actually have IMPACT...percussion. Drums, cynbals, xylophone, or a piano being POUNDED upon don't sound more "impactful" with heavy processing. Instead, they sound softer. It's harder to tell what's being played, or how hard the instrument is being struck.
Lots of people think that "more is better" when it comes to processing. Sadly many of them are mastering engineers who produce today's ridiculously over-compressed/limited/clipped cds. Few of them are musicians! Recordings from decades past sounded MUCH better, because you could hear what the freakin' instruments were playing, and the emotions with which they were playing. Music is about contrast. Without quiet, LOUD HAS NO IMPACT! I've actually heard combinations of heavy processing and low bitrate where instruments DISAPPEAR (finger-cymbals on "Eternal Flame" by the Bangles on "The Shuffle Channel"...the HD2 stream of WLYT Charlotte being an example I've discussed before. The little "dings" which punctuate lines are COMPLETELY ABSENT when they play the damn song! Pre-processing used by HD radio stations (think Neural) is designed to "prepare the audio" to minimize artifacts. HEAVY HANDED PROCESSING INCREASES AUDIBLE ARTIFACTS! The combination of pre-processing and heavy-handed processing can actually punch holes in the music to the point that key elements of the mix disappear. Does that sound "musical" to you?
Make everything (including softly sung vocals) "impactful", and you sacrifice the things which are meant to be impactful...like percussion! In my mixes I do for radio, I DO use compression and limiting to bring things "up front" and make everything sit well in the mix. But I always leave headroom so there's some room for "jump factor"...an explosion, a crash, knuckles sharply rapping on a door, the sharp "yap" of a small dog, these things sound smaller, and LESS "impactful" if you compress/limit the freakin' life out of them. It's the sharp leading edge of a transient, rising above the rest of the mix that says "HELLO...PAY ATTENTION TO ME NOW!"
16 bit digital audio is capable of a maximum dynamic range of 96db (6db pre bit x 16). More than that, actually, as "dither" (low level noise deliberately added to the signal) keeps the "least significant bit" from ever toggling off...which actually makes it possible to reproduce sound BELOW the theoretical -96db point. NO, I'm not suggesting that anybody encode audio that low! Nor am I a purist who believes in no processing. But stations must fine a judicious compromise between having some real dynamics, and consistent average level. Think heavy-handed processing gives more "impact"? Listen to the instruments that actually have IMPACT...percussion. Drums, cynbals, xylophone, or a piano being POUNDED upon don't sound more "impactful" with heavy processing. Instead, they sound softer. It's harder to tell what's being played, or how hard the instrument is being struck.
Lots of people think that "more is better" when it comes to processing. Sadly many of them are mastering engineers who produce today's ridiculously over-compressed/limited/clipped cds. Few of them are musicians! Recordings from decades past sounded MUCH better, because you could hear what the freakin' instruments were playing, and the emotions with which they were playing. Music is about contrast. Without quiet, LOUD HAS NO IMPACT! I've actually heard combinations of heavy processing and low bitrate where instruments DISAPPEAR (finger-cymbals on "Eternal Flame" by the Bangles on "The Shuffle Channel"...the HD2 stream of WLYT Charlotte being an example I've discussed before. The little "dings" which punctuate lines are COMPLETELY ABSENT when they play the damn song! Pre-processing used by HD radio stations (think Neural) is designed to "prepare the audio" to minimize artifacts. HEAVY HANDED PROCESSING INCREASES AUDIBLE ARTIFACTS! The combination of pre-processing and heavy-handed processing can actually punch holes in the music to the point that key elements of the mix disappear. Does that sound "musical" to you?
Make everything (including softly sung vocals) "impactful", and you sacrifice the things which are meant to be impactful...like percussion! In my mixes I do for radio, I DO use compression and limiting to bring things "up front" and make everything sit well in the mix. But I always leave headroom so there's some room for "jump factor"...an explosion, a crash, knuckles sharply rapping on a door, the sharp "yap" of a small dog, these things sound smaller, and LESS "impactful" if you compress/limit the freakin' life out of them. It's the sharp leading edge of a transient, rising above the rest of the mix that says "HELLO...PAY ATTENTION TO ME NOW!"
M
Mike Walker
Guest
As I think about it Len, Hippo and I may not be THAT far apart. He thinks MASSIVE amounts of processing are the way to go. I think considerable processing is necessary, but short of the MASSIVE levels that begin to destroy the "slam" and clarity in a signal. It's not as if I'm advocating wimpy, unprocessed audio. In a radio station, Hippo and I may disagree only on whether to process more, or MUCH MORE!
YIKES! I feel like I just walked on the beach shirtless [ ] without my sunscreen! First, Mike – NOWHERE in my prior post did I call for “MASSIVE” volume compression and on-purpose over-processing... A quick review will reveal the words: “prudent”, “effective”, and “HIGH LEVELS of JUDICIOUS processing”. I’m searching for my Granny Glasses to find any glimpse of my prior support for the “stationary VU needle” syndrome associated with the horrendous [and all too common] audio fallen victim to mischievous processing... Usually, courtesy of some youthful chest-shaving rhythmic-pop PD who managed to get his mitts on the station's digital All-in-Wonder box from one of “The Big Os” : In no way did I advocate that approach!
Remember, there is NO such thing as “headroom” in land of digital audio. Recall the 90s-vintage Sony DAT recorders where the “VU scale” changed from green to red at minus-6VU? That was their digital-definition of “headroom”! Then consider the FACT that distortion [unrelated to “dither”] INCREASES dramatically as digital level falls below minus-10db; and you have described a technical landscape which DEMANDS considerable amplitude control and aesthetic augmentation. Arguably, digital audio is a more-deserving candidate for signal processing than is its analog counterpart.
I will not pretend to be shy about my technical AND personal preference for LOUD and DENSE audio signatures on both AM and FM. I pity those who are so consumed with musical nuance that they end up listening to their audio HARDWARE instead of the program material that utilizes it. The broadcast spectrum is a dirty world – and getting exponentially-dirtier due to IBOC! It bares NO resemblance to a fluffy living room where the $30 Sheffield vinyl plays on an esoteric $2500 turntable/tone-arm/cartridge combo into a rack ‘o Mac and several-grand worth of B&W speakers. Even my fabled McIntosh MR-78 comes NOWHERE CLOSE to your recollection of a pristine 96db dynamic range... That's a laboratory measurement, Mike – NOT a receivable reality! I can’t speak for you folks, but I prefer an altered state when I’m facing that landscape known as a root-canal in the dentist’s chair... Listeners on Wally World radios likely feel the same about demodulated audio!
Realizing the total lack of evidence that compression ratios have any relationship to radio ratings, I’ll have to conclude that the audience cares little – and generally-favors the station that is easiest-heard above the roar. Here’s a story: My staff and I were once identified and accosted by a listener of my FM classic hits station at an Applebee’s Happy Hour. His pants were pulled tight over the issue of our somewhat-aggressive transmission of Sugerloaf’s “Green-Eyed Lady” [you’ll recall that song’s signature staccato/pause x5]. He complained about “that sucking sound” between each riff... Then provided a live imitation of our air-chain between brews number two and three. A thirty-something Joe Onlooker at a neighboring table stood up and shouted: “HEY... I like that song and that station... Get a life, DUDE!”
As for the newest excuse to argue – known as “How to process before a CODEC”... I learned long-ago to avoid all the processor puff-pieces written by their purveyors – ESPECIALLY Omnia and Orban. I generally laugh at their invention of that “Magical Mystery Tour” called “Neural pre-processing for low-bitrate streaming”—merely a gimmick to incite “must-have” status for their box when some innocent broadcast wanna-be hops onto the internet with his ShoutCast server and limited bandwidth!
The effective solution is far-more simple – albeit less complimentary to Orban’s bank account. Understand first: While a lower-bitrate CODEC may offer an illusionary 32 or 24kHz sample rate; you are better-off facing reality and choosing a lower-octane flavor [20 or 16k]. That's a guaranteed strike-two on the fizzie artifacts. If the available pipe and user target limits your stream to a bit-reduced 10kHz audio bandwidth, and you wish to elevate the 5-8kHz arena to compensate [and create a brighter perception] – DON’T jam the encoder with jacked 8-15kHz content! Here’s but one toil-acquired tool I’ll share to minimize the dreaded “chorus effect” – and it costs only $250... The Rane PE-15 five-band parametric EQ with high and low-pass shelving in each extreme band. Band Five [the highest], actually offers a selectable cut-off spanning 1kHz to 20kHz with a variable “roll” from pastoral to cliff. Turn it on at 8kHz and instruct it to dive – dive – dive! Then visit eBay [or your local used guitar shop] and pick up a clean decade-old DBX 166A for about $100... You’ll be pleasantly-surprised that you’ve just managed to out-perform EVERY CCU station stream from your bedroom 8)
] without my sunscreen! First, Mike – NOWHERE in my prior post did I call for “MASSIVE” volume compression and on-purpose over-processing... A quick review will reveal the words: “prudent”, “effective”, and “HIGH LEVELS of JUDICIOUS processing”. I’m searching for my Granny Glasses to find any glimpse of my prior support for the “stationary VU needle” syndrome associated with the horrendous [and all too common] audio fallen victim to mischievous processing... Usually, courtesy of some youthful chest-shaving rhythmic-pop PD who managed to get his mitts on the station's digital All-in-Wonder box from one of “The Big Os” : In no way did I advocate that approach! Remember, there is NO such thing as “headroom” in land of digital audio. Recall the 90s-vintage Sony DAT recorders where the “VU scale” changed from green to red at minus-6VU? That was their digital-definition of “headroom”! Then consider the FACT that distortion [unrelated to “dither”] INCREASES dramatically as digital level falls below minus-10db; and you have described a technical landscape which DEMANDS considerable amplitude control and aesthetic augmentation. Arguably, digital audio is a more-deserving candidate for signal processing than is its analog counterpart.
I will not pretend to be shy about my technical AND personal preference for LOUD and DENSE audio signatures on both AM and FM. I pity those who are so consumed with musical nuance that they end up listening to their audio HARDWARE instead of the program material that utilizes it. The broadcast spectrum is a dirty world – and getting exponentially-dirtier due to IBOC! It bares NO resemblance to a fluffy living room where the $30 Sheffield vinyl plays on an esoteric $2500 turntable/tone-arm/cartridge combo into a rack ‘o Mac and several-grand worth of B&W speakers. Even my fabled McIntosh MR-78 comes NOWHERE CLOSE to your recollection of a pristine 96db dynamic range... That's a laboratory measurement, Mike – NOT a receivable reality! I can’t speak for you folks, but I prefer an altered state when I’m facing that landscape known as a root-canal in the dentist’s chair... Listeners on Wally World radios likely feel the same about demodulated audio!
Realizing the total lack of evidence that compression ratios have any relationship to radio ratings, I’ll have to conclude that the audience cares little – and generally-favors the station that is easiest-heard above the roar. Here’s a story: My staff and I were once identified and accosted by a listener of my FM classic hits station at an Applebee’s Happy Hour. His pants were pulled tight over the issue of our somewhat-aggressive transmission of Sugerloaf’s “Green-Eyed Lady” [you’ll recall that song’s signature staccato/pause x5]. He complained about “that sucking sound” between each riff... Then provided a live imitation of our air-chain between brews number two and three. A thirty-something Joe Onlooker at a neighboring table stood up and shouted: “HEY... I like that song and that station... Get a life, DUDE!”
As for the newest excuse to argue – known as “How to process before a CODEC”... I learned long-ago to avoid all the processor puff-pieces written by their purveyors – ESPECIALLY Omnia and Orban. I generally laugh at their invention of that “Magical Mystery Tour” called “Neural pre-processing for low-bitrate streaming”—merely a gimmick to incite “must-have” status for their box when some innocent broadcast wanna-be hops onto the internet with his ShoutCast server and limited bandwidth!
The effective solution is far-more simple – albeit less complimentary to Orban’s bank account. Understand first: While a lower-bitrate CODEC may offer an illusionary 32 or 24kHz sample rate; you are better-off facing reality and choosing a lower-octane flavor [20 or 16k]. That's a guaranteed strike-two on the fizzie artifacts. If the available pipe and user target limits your stream to a bit-reduced 10kHz audio bandwidth, and you wish to elevate the 5-8kHz arena to compensate [and create a brighter perception] – DON’T jam the encoder with jacked 8-15kHz content! Here’s but one toil-acquired tool I’ll share to minimize the dreaded “chorus effect” – and it costs only $250... The Rane PE-15 five-band parametric EQ with high and low-pass shelving in each extreme band. Band Five [the highest], actually offers a selectable cut-off spanning 1kHz to 20kHz with a variable “roll” from pastoral to cliff. Turn it on at 8kHz and instruct it to dive – dive – dive! Then visit eBay [or your local used guitar shop] and pick up a clean decade-old DBX 166A for about $100... You’ll be pleasantly-surprised that you’ve just managed to out-perform EVERY CCU station stream from your bedroom 8)
M
Mike Walker
Guest
I actually agree with all of the above, Hippo, and confess MY preference for loud, dense audio. I still LOVE the "gallon in a quart container" sound of densely processed music on an AM station, received on a good radio. I suppose I should have confined my comments to "even if you love processing, it's possible to go too far".
I think what you're describing about not jamming too much hf content into a low bitrate stream is largely the same point that I've made here (and elsewhere), that it's better to restrict high frequency content if the bitrate is low. I suspect that's largely what "pre-processors" (such as Neural) do dynamically, moment to moment, is clip those high level, extreme highs at the moment where artifacts would result...hence the missing "finger cymbals" on WLYT-HD2 when they play "Eternal Flame". Limiting highs with a sharp 9 or 10khz low pass filter prior to the processor would probably accomplish the same thing (prevention of artifacts), without the side effect of removing percussive instruments. Perhaps we don't disagree that much at all!
I think what you're describing about not jamming too much hf content into a low bitrate stream is largely the same point that I've made here (and elsewhere), that it's better to restrict high frequency content if the bitrate is low. I suspect that's largely what "pre-processors" (such as Neural) do dynamically, moment to moment, is clip those high level, extreme highs at the moment where artifacts would result...hence the missing "finger cymbals" on WLYT-HD2 when they play "Eternal Flame". Limiting highs with a sharp 9 or 10khz low pass filter prior to the processor would probably accomplish the same thing (prevention of artifacts), without the side effect of removing percussive instruments. Perhaps we don't disagree that much at all!
On HD radio, isn't everything above 7KHz 'fake' band replication anyhow?
M
Mike Walker
Guest
The frequency at which spectral replicaton begins varies (upward) with increasing bitrate. But yes, at lowbitrates, it starts quite low, which is why I suggest a lowpass filter set for 9khz or so for low bitrate streams. I guarantee listeners will never say "where are the highs?", because the vast majority of content is below 10khz anyway. And it should minimize artifacts by allowing the codec to concentrate it's energy on the critical midrange, where the ear is most sensitive.
Spectral replication can be quite accurate sounding and convincing, BECAUSE there are no fundamental tones up there. What's above 10khz is harmonically related overtones, which actually are pretty much matematically predictable based upon what's happening "down below". It works quite well. If you're an XM subscriber, you've been hearing "spectral replication" for years, and some of their channels sound superb (others slightly lower in quality than two tin-cans and a tight string).
Spectral replication can be quite accurate sounding and convincing, BECAUSE there are no fundamental tones up there. What's above 10khz is harmonically related overtones, which actually are pretty much matematically predictable based upon what's happening "down below". It works quite well. If you're an XM subscriber, you've been hearing "spectral replication" for years, and some of their channels sound superb (others slightly lower in quality than two tin-cans and a tight string).
- Status