Audio chain conversions between digital and analog

feedback.pdxradio.com forums feedback.pdxradio.com forums Portland Radio Audio chain conversions between digital and analog

This topic contains 13 replies, has 8 voices, and was last updated by  lastday 6 months, 3 weeks ago.

Viewing 14 posts - 1 through 14 (of 14 total)
  • Author
    Posts
  • #41307

    lastday
    Participant

    Assume a station’s automation system is populated with 320 kbps MP3s @ 48000 or 44100 sample rate. Basically the best quality available in the lossy MP3 format.

    When played back, does the automation system do D-A conversion on the MP3s, handing off analog audio to the next hop in the chain? Or does it hand off the MP3 in its digital form?

    After the audio leaves the automation server, on its way through processing and eventually to the transmitter, is it ever converted back to digital using some form of lossy compression? Or is all conversion done in lossless form, like 16/44.1 “red book” CD quality (or better).

    Eventual lossy conversion for HD is not part of the scenario.

    I’m asking on behalf of a friend. 😉

    #41308

    Andy Brown
    Participant

    There is no unique answer because there are so many options available, depending primarily on budget.* Throw in digital STLs and there is another A/D and D/A conversion that might happen.

    Bottom line: Every A/D and D/A conversion chip or circuit is not created equal. That is to say, the more conversions that take place, the greater the chance for noticeable degradation in the program audio chain. This is why there are now digital consoles, digital STLs, and transmitters and exciters that accept digital input streams. Also, D/D conversions are no bargain, either. Best practices is to minimize the conversions and utilize quality converters.

    *Budget not just for the hardware, but for STL, internet and intranet bandwidth. The more robust the digital stream, the more bandwidth it needs.

    #41316

    theedger
    Participant

    Who plays lossy audio as their source material on the radio? This is 2019, storage space is very inexpensive. Crap in = crap out. A tip for a friend.

    #41317

    Andy Brown
    Participant

    Most LPFMs do.

    Analog FM only can pass 15k, so MP3 or AAC 256 kbps or greater is sufficient.

    And the larger limitation is that most LPFMs can not use PCM on their digital STL Barix codecs because the bandwidth that it requires is too costly. So even if they’re playing AIF or WAV uncompressed digital files or vinyl, the STL ends up using some compressed scheme.

    It has nothing to do with storage space.

    #41319

    Screamer
    Participant

    Not to mention the fact that some automation systems can only handle 44.1 or 48k sample rates regardless of the wrapper. *cough* Simian *cough*

    #41320

    QPatrickEdwards
    Participant

    For our LPFM, all music files are uncompressed 44.1 KHz PCM, while all of our block programming that we download via FTP ranges from 192-256 kbps mp2 and mp3 all the way down to 64k mono mp3.

    We just recently installed a Barix pair to stream our program audio from our studio in McMinnville to our transmitter out in the sticks (using stereo 384kbps mp3 stream) to replace the pair of old Dell Poweredge servers that we used to use to send programming at 192kbps mono .ogg streaming, using a local Icecast server. Not elegant, but it worked well enough for the time we needed it.

    We would use uncompressed PCM to the transmitter, but the wireless connection at the transmitter end isn’t robust enough to support it and be reliable.

    The audio from our automation is analog through our board, into our EAS, through some audio limiting and out to the Barix. At the transmitter audio flows in analog from the outputs of the Barix to our audio processor/stereo generator. I still need to find a good basic multi-band analog audio processor to pre-process the audio out at the transmitter site…but that is another story for another day.

    #41346

    lastday
    Participant

    Here are two versions of The Ending by Tedeschi Trucks Band. One is an uncompressed 16/44100 WAV file (50MB), the other a lossy MP3 at 320 kbps 48000 (11MB). The master recording they were made from was ripped from vinyl at 24/96000 (164MB). Some noise reduction was applied and a couple clicks repaired but it’s still obvious this is from vinyl.

    https://drive.google.com/open?id=1DBlAH7zbgW3ysduU0IsdIR0-tMmHO9lV

    Can you hear any difference between them?

    If an MP3 of that quality doesn’t go through any more rounds of lossy compression on its way to the transmitter, what would be the point of using orders of magnitude more drive storage space for the WAV versions?

    OTOH, if there’s more lossy compression in the audio chain, a good argument can be made for using the CD-quality WAV version.

    #41347

    Andy Brown
    Participant

    First off, using the open internet to compare two audio files is a fool’s errand, but even my old, battered, tinnitus distorted ears can hear the difference. However, that is not a valid experiment. No, not in the least.

    Not only is a desktop, laptop or phone being fed from wifi being fed by the open internet a problem for qualitative comparison but also a single piece of music is never sufficient to draw any meaningful conclusion.

    Having said that, let’s be for real. This isn’t about storage space. It’s about the cost of equipment and bandwidth. In commercial radio, fiber optic lines are leased and contain only the program audio chain material and supporting data for RDS, control and telemetry. In LPFM radio, the open internet is usually the medium of choice (unless the studio and transmitter are close enough to use a new age RF STL or if it’s really close, a piece of wire). In LPFM radio, there is a constant struggle between those that know what they are doing and those that are paying the bills as to what is going to be done. It’s politics and not performance that often prevail. There is a lot of misinformation that many hold as fact about acoustics.

    As an old dog at this game, a lot depends on the kind of audience listening environment you are playing to. When FM overtook AM, it was all about the mobile audience. The mobile listening environment is extremely noisy. The 40 year crowding of the FM dial has made the amounts of ACI and THD the main factor, not the freakin’ source material.

    These days, streaming and digital issues like dropouts and in general the overall BER (bit error rate) plague HD radio reception even though it’s more immune to multipath. In LPFM (analog) radio, just like commercial radio of the past, the limited bandwidth, multipath and the ACI play a more pivotal role than whether the source is a higher quality MP3/AAC or a fully robust PCM WAV/AIF or even an analog direct from vinyl into the transmitter. Also, in analog FM there is always (like on vinyl) a pre-emphasis curve applied before transmitting/pressing and in broadcast a four step processing regimen (equalization, compression, expansion, limiting) going on in most program audio chains because without it, you will sound weak and distant.

    So you see, you are splitting hairs. In the final analysis it comes down to frequency response of the input being robust enough to pass through the chain with a minimum of damage. Radio is not like going to Fred’s Sound Of Music and listening in a sound room to your favorite album on big speakers and a spendy amplifier regardless of whether the source is a CD or vinyl.

    It’s important to remember that how well a radio signal can deliver quality audio is not dependent on any one factor previously touched on, it’s a combination of all of them plus the quality of the RF signal being detected at the receiver, the quality of the receiver, and the quality of the listening environment.

    In the end, in the world of LPFM where DJs bring in their laptops and plug into the program audio chain which may or may not have been designed, wired and maintained by somebody that knows what they are doing, your question has little relevance. Sure, if I could have 20,000 WAV files on my laptop I’d do it, but we aren’t there yet. Besides, audiophiles never looked to radio for fidelity and radio listeners shouldn’t be looking for purer source files to solve problems of listening quality.

    You wrote “If an MP3 of that quality doesn’t go through any more rounds of lossy compression on its way to the transmitter” and that pretty much points to the issue, i.e. it will (go through more digital compression) and it will also go through further analog processing so it can be heard on an overcrowded dial. FM is not the place for purists of any kind. A balanced solution can be had that will apply processing where it is needed so that the best sounding sources aren’t highly degraded and the poorer ones are given the help they need.

    #41348

    Alfredo_T
    Participant

    A number of programs on XRay FM sound as though the hosts have harvested some of the tracks they play from the Internet. The codec artifacts are noticeable, even when listening on a car radio. Thus, I would say that in the world of non-commercial volunteer radio that doesn’t implement playlists or similar restrictions, it being 2019 seems to facilitate the presentation of compressed audio on the air.

    #41350

    lastday
    Participant

    It’s kind of like being on the receiving end of a lecture by Professor Kingsfield (the John Houseman version). 😉

    My foolish question is rooted in trying to understand why there are such wild variations in audio quality from the same station, depending on the playback source. Like why some 256 MP3s sound fine and some are unlistenable, as if they’ve been through 5 rounds of lossy compression. Also surprised at how good vinyl playback sounds on their new turntables.

    I too have tinnitus but it doesn’t prevent me from identifying good vs crappy audio. Most of the time.

    #41351

    Andy Brown
    Participant

    “why there are such wild variations in audio quality from the same station”

    Two basic reasons:

    1. The original source quality can vary widely. Some files are original rips and others are probably from downloads while even others might be downconverted from lossless digital files. Taking it even a step further, some tracks from old CDs were mastered poorly in only 16 bit and not normalized. Twenty four bit 96 kHz normalized masters make much better 16 bit 44.1 kHz CDs and any compressed format down converted compressed digital file will be improved as a result.

    2. Not all A/D converters (chips and/or circuits) are of the same quality, so not all 256 MP3 files are being created equally regardless of how pristine the source is.

    I totally get it, lastday. There just is no easy solution that is implementable unless you can control the creation of every source.

    #41356

    Steve Naganuma
    Participant

    Maybe not directly in the audio chain but one of my favorite audio plugins is the Waves C4 Multiband Compressor. It also has a UNcompressor feature which does a great job at cleaning up overly compressed audio files before putting them in a digital on-air system. Here is a video showing some of what the C4 plugin can do.

    #41359

    radiogeek
    Participant

    Funny, I just posted. And it didn’t appear.

    So I turn off my VPN and see what happens.

    There you go, the forum won’t allow postings via a VPN. What a hassle.

    Codecs aren’t bad unless you keep changing them and re-saving them. Re-sampling is what creates the crap we don’t want to listen to.

    #41361

    radiogeek
    Participant

    “why there are such wild variations in audio quality from the same station, depending on the playback source. Like why some 256 MP3s sound fine and some are unlistenable, as if they’ve been through 5 rounds of lossy compression. Also surprised at how good vinyl playback sounds on their new turntables.”

    Exactly. It’s not the bandwidth of the codec, it’s how many times it’s been f*(%$d with. You can have 256kb of crap, just like a trying to print an 8×10 color print from a 72dpi image that looked sharp online.

    Always go to the source file, preferably uncompressed for ALL editing steps. Only apply a codec at the last moment, and only apply the codec that makes sense for the ultimate use.

    As I’ve left radio, the only audio editing I’ve done lately is for a non profit website, spoken word. I have a clean recording with good signal to noise ratio, get the gain right and do some light peak compression, then save at 96kb MONO and send to the org to post on their website. Sounds clear as a bell and folks can’t stop telling me how happy they are. Why? It’s for streaming online so the low bit rate means no buffering, even if you’re on wifi at PDX. That I went directly from good uncompressed to the final use codec in one pass means I haven’t accumulated the crap that re-encoding creates.

    The key is to have an un-corrupted source file to start with. As Andy said, you don’t always have this necessary control over your sources.

Viewing 14 posts - 1 through 14 (of 14 total)

You must be logged in to reply to this topic.