May 21, 2020 at 12:36 pm #46430
For my amusement, I did some comparison between one of the original WAV files, and the same thing captured from KRVM-HD1. I normalized the “aircheck” to -6dB so it would be the same peak level as the original file (don’t yell at me). When I have more time I’ll see about posting the audio samples and screenshots from Audacity.
The “aircheck” was from a Sony XDR-F1HD that’s been heavily modified to correct the original drooping high frequency response and improve things like S/N.
Their Left & Right channels are reversed.
Their HD1 frequency response does extend out to 20kHz. I know for sure that when KLCC was in HD it had nothing above 15kHz.
The dynamic range of the air check sample is nothing like the original. Not unexpected.
In terms of spectral content, they look surprisingly similar, considering everything the audio has probably been through.May 21, 2020 at 1:25 pm #46431Alfredo_TParticipant
I had a little bit of a chuckle regarding the transposed left and right channels, as I would have otherwise mistakenly assumed that in the modern world of digital audio, this should not happen. On the other hand, the signal chain at KRVM or anyplace else is a black box.
I think that the request for -6dB normalization has to do with how the gain is set at various points in the airchain, and it is the modern equivalent of telling DJs to not allow the VU meters on the console to go over 0 dB. Years ago, I met a DJ from a volunteer station who believed that running the meters “in the red” would drastically shorten the life of the mixing console. At my university’s station, the training materials contained a warning that allowing program levels to exceed 0 dB would cause the station to sound as though it were being played through cheap “walkman speakers.” In reality, running the levels slightly on the hot side would cause the processing to engage more, if you will, so that there would be less audible difference between soft and loud passages. In effect, this would cause these volunteer stations, which are very likely trying to avoid the trappings of commercial radio and appeal to more serious music aficionados, to sound more like contemporary hit music stations. Telling the DJs that hot levels are bad is a behavior-altering technique.May 21, 2020 at 1:44 pm #46432nosignalallnoiseParticipant
I had a little bit of a chuckle regarding the transposed left and right channels, as I would have otherwise mistakenly assumed that in the modern world of digital audio, this should not happen.
Well, Muzak’s stereo simulcast of “Jukebox Gold” on Echostar had the channels swapped for years, and probably still does for all I know (I haven’t airchecked it in well over a decade). Granted it’s not technically broadcast radio in the traditional sense, even though it’s produced using basically the same technology.
I think that the request for -6dB normalization has to do with how the gain is set at various points in the airchain, and it is the modern equivalent of telling DJs to not allow the VU meters on the console to go over 0 dB. (R)unning the levels slightly on the hot side would cause the processing to engage more, if you will, so that there would be less audible difference between soft and loud passages. In effect, this would cause these volunteer stations, which are very likely trying to avoid the trappings of commercial radio and appeal to more serious music aficionados, to sound more like contemporary hit music stations.
I believe that would be correct. Similar principles apply to analogue tape, where 0 VU is the theoretical “maximum” level that most common magnetic emulsions (i.e. ferrics) are able to record full dynamic range without going into saturation and clipping.May 21, 2020 at 2:11 pm #46434
Well I tried posting screenshots and links to the audio clips but the post has vanished.
Back before HDTV, Comcast in Eugene had the audio channels transposed on the FOX station for years. A car would drive by left to right but the audio would pan right to left. OTA the channels were not transposed.
Trying the audio clips again (stereo channels deliberately swapped on the aircheck to match the original audio). The air check sounds louder presumably due to their audio processing (you tell me!). Both clips are normalized to the same peak level but with no other deliberate manipulation. That air check audio actually sounds pretty good considering everything it’s been through. Probably had least three rounds of lossy digital compression applied, plus the station’s audio processor.
Original uncompressed audio: https://1drv.ms/u/s!Aio5y5k9Z_O-sDhllFMpggavjA-L?e=DBXbGb
Aircheck audio: https://1drv.ms/u/s!Aio5y5k9Z_O-sDn2TzvDLXjRsd-2?e=VvWbWq
May 21, 2020 at 2:41 pm #46436
- This reply was modified 1 year, 2 months ago by lastday.
I just remembered something… When KRVM first deployed HD, the analog and digital programs had the channels reversed from each other. Analog left was digital right and vice-versa. It was obvious in a car when driving in places distant from the station that caused the receiver to switch in and out of HD. I told them and the explanation I got was the exporter needed a firmware upgrade. Whatever, they did get the channels matched on analog and digital, but apparently now both analog and digital are reversed.
Not that it makes any difference in the real world. I never noticed it until I started messing with this OCD project.May 21, 2020 at 4:03 pm #46438Jeffrey KoppParticipant
This may be off-topic, but I use Windows Media Player to rip my CDs to .mp3 and make CDs of them, 10-14 CDs on each, to put in my 6-CD changer. I have my whole collection loaded at once.May 21, 2020 at 4:27 pm #46439
I’ve derailed my own thread slightly. It’s all good and kind of loosely related.May 21, 2020 at 5:50 pm #46441Andy BrownParticipant
“Not that it makes any difference in the real world”
Not so much in radio, but in TV or video, it does. If the car appears on the left and the sound ramps in from the right . . .
In fact, it kind of does matter in audio, too. It doesn’t degrade quality any as long as the phase of one channel doesn’t get reversed, but it does effect stereo imaging and the biggest danger is audience alienation when a certain instrument solo is on the opposite side from the way they have been listening to it at home and on the radio for years.
You make it sound like the station has been shoddily engineered and needs to be gone through from stem to stern.
A few words about 0 VU. It’s not the same from station to station. The pure definition of 0 VU is +4 dBu (1.228 volts RMS) but in an analog pro audio balanced line environment generally means 0 dBm (1 milliwatt across a 600 Ω line). Remember, those kinds of numbers are when using a single frequency tone, not music.
Also, most pro audio equipment won’t actually clip until you stuff far beyond the 0 level on its input meter, like up over 20 dBu or even 28 dBu.
The most important difference to note is that there is a huge difference between 0 VU on a piece of consumer gear where it means -10 dBV and in pro audio gear where it means +4 dBu or +4 dBm. Also, the source and load impedances are all different. Plug a high Z consumer piece into a pro audio low Z 600 Ω line or 150 Ω mic input and the consumer piece essentially sees a short, it’s looking for a 10 kΩ load, not 600 Ω or 150 Ω.
When I worked at KATU I discovered their entire (mono at first, later stereo) audio chain was set up for 0 VU to be +8 dBm. So again, not all standards are really universal. BTW, broadcast consoles can usually be set up for 0 VU to be anywhere between -10 and +28 dB. It comes down to how much headroom do you want for distracted operators who don’t pay attention to the meters anyway. They (DJs) always got depressed when I told them it didn’t really matter how they set the levels since the processing would suck up what’s low and crunch down what’s too loud much to their disappointment. They would say stupid stuff like “I want that part of the song to be really loud!” Sorry, dude, you can punch it all you want, it won’t matter. Of course, if they really overdrive it the sound quality goes to hell because of clipping, but you’d be surprised how tolerant a well balanced program audio chain can be. One thing for sure, using a source normalized to -6 dB is a mistake, and although it probably is a workaround of some sort, clearly indicates something is not set up correctly down the line. Period.
https://en.wikipedia.org/wiki/VU_meter (careful, some of what this link says is not exactly correct).May 21, 2020 at 6:27 pm #46443
I should have said “Makes no difference for radio”. It absolutely does for video. I consider myself picky with regard to audio and I never noticed the channels were swapped. But I haven’t had an “audiophile” system since around 1990.
We haven’t even gotten into phono cartridges and phono preamp impedance loading. It makes a difference. Most cartridges spec a load of 47K ohms. This is not necessarily always correct.May 22, 2020 at 3:15 pm #46456
It appears that using the Normalize function in Audacity can be problematic with audio sourced from vinyl. If there’s even a single loud “tick”, that’s the peak level used by Normalize to adjust the level of the entire track. That single tick may be at -1dB in the raw file. Normalizing to (say) -6dB will reduce the level of everything by 5dB. That’s what it looks like. ?
I found a plugin for Audacity called ReplayGain. It seems to better handle the random loud “tick” without drastically affecting everything.
“ReplayGain is different from peak normalization. Peak normalization merely ensures that the peak amplitude reaches a certain level. This does not ensure equal loudness. The ReplayGain technique measures the effective power of the waveform (i.e. the RMS power after applying an “equal loudness contour”), and then adjusts the amplitude of the waveform accordingly. The result is that Replay Gained waveforms are usually more uniformly amplified than peak-normalized waveforms.”
And the latest version of Audacity 2.4.1 has a new Effect called “Loudness Normalization” which seems to work better with vinyl.
“Why use Loudness Normalization rather than Normalize or Amplify?
Using Loudness Normalization rather than Normalize or Amplify helps you more easily set the required LUFS loudness target when normalizing.
When preparing audio for television/radio programmes, podcasts and some websites you may be subject to loudness restrictions for the audio. Loudness is usually measure in LUFS (Loudness Units Full Scale). The LUFS level restrictions can vary by application. For example the level for television in the US is normally -24 LUFS and the EBU (European Broadcasting Union) recommends -23 LUFS. Out of all the standards, this one is the most serious in that a television network can get its broadcast license revoked for a violation. Send in a program with a higher level, and it will be returned for a revision.
Another use case is creating an equally loud playlist from different sources.”
That took me to LUFS (LKFS).
“Radio shares many issues with television; however, radio has unique complexities around transmission and listener practice. While adoption of a global standard within radio is still a little way off, much research into loudness normalization has already been undertaken and, in many cases, the practice is already well-established.”May 22, 2020 at 9:44 pm #46472nosignalallnoiseParticipant
I found a plugin for Audacity called ReplayGain. It seems to better handle the random loud “tick” without drastically affecting everything.
Page doesn’t mention a word about Audacity. Is there a different download link for the plugin?May 23, 2020 at 12:52 am #46473Andy BrownParticipant
Generally you edit out the pops and ticks before you normalize something. There are multiple ways of doing that but normalizing should be your last step. Normalizing isn’t even really part of ripping. Have you spent any time with https://manual.audacityteam.org/man/sample_workflow_for_lp_digitization.html ???
Your track by track approach is, to me, the long way home. I create a three hour show by recording it all in one session, just like I was doing it live (almost exactly with a few caveats). Then I go through the one file that is created in Audacity while it’s in the best form it could be (32 bit/actually 24 bit). I raise up the songs that came through low level with the amplify tool, edit out the mistakes if possible, edit some of my breaks to get it to 180 minutes as close as possible without sacrificing too much dialogue, and then normalize. The important thing is not to overdrive anything when the recording is being made because once anything gets clipped, you can reduce the gain after the fact but you cannot eliminate the clip distortion at the flat topped peak. Sometimes you can hear it, sometimes not. Mostly I need a hardware limiter for my vocals and for the mixer output. I’m thinking about getting one of those old dbX 1046 units (discontinued). Then I could run everything a little hotter without worrying about the unfixable clipped flat peaks. Anyway, I’ve been sending an MP3 320 kbs export of the show and it sounds great. It drops right into the automation like it was a three hour song. In your situation, you would eliminate the need for anyone to assemble the show and subsequently screw it up. Also, because I like to play some content that is inappropriate for my time slot I use Audacity to edit out (with silence or the reverse filter) those tracks in advance of the show. In those cases it involves importing the track from my library where it is stored in m4a or mp3, performing the edits, then exporting back to the same format and frankly, they sound just as good as the unedited track I started with.
I realize that you would still have to create the digital tracks from vinyl unless you have two turntables and a microphone. Now that would be an old school recording without needing any tape.
Good luck. I doubt you’ll ever get a total handle on all your quality issues until you find why they insist on normalizing at half quality.May 23, 2020 at 7:31 am #46475
Page doesn’t mention a word about Audacity. Is there a different download link for the plugin?
Search the board for “replaygain” (one word). Look for the thread titled simply “ReplayGain”. The link to the plugin is near the bottom of the first post in that thread as “NEW VERSION”. The plugin is “replaygain.ny”.May 23, 2020 at 7:58 am #46476
Yes, I’ve read and applied their recommended workflow for digitizing LPs for my own use. It’s much faster and more convenient to do things that way, especially when working with entire albums.
In this case the station requested individual files with names in the format “Artist – Track – Vinyl.mp3”. They also declined my offer to string multiple files together into say 15-20 minute sets. They want to control that. Something to do with how they time and insert breaks into Simian. I don’t know the details.
As far as fixing all the pops and clicks, well the show is Vinyl Revival. Anyone tuning in specifically for this show wants to hear vinyl, warts and all. It’s part of the character of the format. I do fix the occasional flaw but I try to keep that to a minimum. If everything is “perfect” then there’s no point.
I’m sure if we had more time to collaborate we could work out a better workflow for everyone, but this situation (can’t play vinyl live) is hopefully temporary. I did mention to them at the get-go that this process is probably not sustainable over the long term because it’s just so time consuming.
But I’m still in for now. 🙂May 23, 2020 at 10:37 am #46477
If the goal is to repair as many flaws in the digitized vinyl as possible, ClickRepair by a guy named Brian Davies produces amazing results. He also has a de-noise tool. They’re not free or very user-friendly, but they’re far more advanced than the tools provided with Audacity.
- You must be logged in to reply to this topic.