What is Mastering and a what does a mastering engineer do

Started by Elantric, July 30, 2012, 02:09:18 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

Elantric

What is Mastering and a what does a mastering engineer do

This Video explains:

musicman65

Ok....is this a quiz? .. I'll bite.

A Mastering Engineer works with the final mix down usually as a 2 track stereo format in audio applications. His goal is to set the overall level and headroom of the track in relation to the maximum level possible.  Often compressors/limiters are used.  They also adjust the overall eq for best sound during playback. This is usually necessary if the recording studio's raw mix is not balanced.

If you don't think your recordings can benefit from mastering,  download a trail of Har-Bal software and open a mix down with it. It is an unlimited parametric eq with visualization software that is amazing. I use it to correct the funky resonances of live recordings.

Anyone else want to take a cracks at it?

bd


Brent Flash

Well, being a retired mastering engineer, I see the title as a little more than compressing songs to match levels. The title started out as the engineer that made the master for replication of records. As the trade evolved mastering engineers found that they could manipulate the width of the grooves in the record master to make more of less material fit on one side. The trade off of this changed the amount of bass mostly so you had to be careful what you did. Spacing between songs, how it started, how it ended were also things the mastering engineer did to the master before it was pressed into records.

As the trade evolved more and more could be done in the mastering process but it is still more or less the engineer that makes the final master that CDs (for example) are made from.

shawnb

I found this post very, very helpful over on cakewalk.com.  I've kept a copy printed out, & I've annotated it here & there based on my few mastering jobs.  This is a very simple, very practical guide for a home studio mastering job from "Yep" on the cakewalk forum:
http://forum.cakewalk.com/tm.aspx?m=475013&high=mastering+tips

Text copied in-line:

QuoteDon't master your own mixes. If the mix doesn't sound good enough to be called a master, then fix the mix. Mastering engineers are people who specialize in doing what the mix engineer cannot do. They make corrections to compensate for inadequacies in the mixing engineer's listening environment, monitoring setup, hearing ability, and so on.

If you can't afford to send your mixes to a mastering engineer, then there a couple of very basic things you can do on your own to make your CD a little more listener-friendly:

Bounce all the songs to 24-bit stereo clips, making absolutely sure that you have no digital overs on the main outs. Do not include fade ins/fade outs in the mix. Just leave the intro/outro noise and silence in the track. We'll fix those in the mastering stage.

Determine which song is the best-mixed, and which translates best on different sound systems. Friends can be helpful in this regard. Your friends and family may give you skewed advice on the quality of your material (some will say that everything is brilliant, others find fault with anything), but if you ask them which one sounds the best-recorded or most professional, they will probably give you pretty consistent answers. Keep track of which one is the best-mixed, since we will use it later as a reference.

Home mastering Stage 1 Here, we are ONLY concerned about how the songs sound in relation to one another-- ignore everything else:
Load all the tracks in the order you think you want them to appear on the record, and put them in one stereo track of one project in your software, but keep them as separate clips of audio. Space them far enough apart that you will have room to move them back and forth a little. Set the master out of your software low enough so that you have plenty of headroom, maybe like -12~18dB or so-- don't worry about the level just yet, just make sure that you have it quiet enough that you are going to hear the whole thing with no clipping, either at the converters, your amplifier, or the speakers, even if you have turn things up here and there.

Now turn the level of your monitors up to a normal listening volume (RMS level of 83dB SPL is industry-standard, about as loud as city traffic or a noisy restaurant). Play them all back and listen, focusing on the transitions between songs, and A/Bing each song against the "best mix" frequently. We are going to make some very basic adjustments to try and get them all to sound good in comparison with each other. When in doubt, use the above-selected "best mix" as a reference. The other songs should compare favorably in an A/B test with that one. Do NOT, at this stage, compare any individual song with a commercially-mastered CD. We'll get to that later.

Overall Volume:
If a ballad seems louder than a rocker, then turn down the ballad or turn up the rocker 'till they seem proportionate. Use clip envelopes for this. Use your "best mix" as a reference, turning the other songs up or down so that they seem proportionate to the best mix and also to the songs before and after one another. Don't sweat the levels or digital overs or anything like that just yet, just get all the tracks so they seem to have the correct proportional loudness, as though they were being played in order by an actual band. And use the best mix as your reference point. Don't change it's volume, change the volume of the other songs to suit it. And go by ear, not by your meters.

Instrument levels:
If the vocal sounds suddenly huge and dominating on one track compared with the best mix, then go back to the mix and lower the vocal track by a few dB until it sounds proportional. If the kick drum seems to disappear, go back to the mix and raise it a couple dB. Make sure to save these "remixes" as separate projects from the original mix (something like: "Minimum Rage-vocals down 3dB"), unless you're positive that you're making changes that you will always want to keep, even when you can afford a million-dollar mastering engineer.

Frequency balance:
Does one song appear to be really bottom-heavy, or another seems tinny, or maybe to have too much presence and not enough highs and lows compared with your best mix? If so, then you should probably remix it. If you are ABSOLUTELY positive that the mix is good, but that it just happens to be a little bottom-heavy or whatever, then you are allowed to use a tiny bit of corrective equalization to balance it out with the rest of the album. Use the best equalizer you own. You are only allowed to use the low cut/boost and high cut/boost filters, and no more than 3dB of either one, with a gradual Q (let's say 1 or lower). Any more than that, and you have to remix. Sorry, those are the rules. You are permitted to use ONE AND ONLY one cut or boost of up to 6dB per album, but your overall score drops one letter grade if you use it.

Spaces in between songs/tuck-and-tail:
Drag the song clips back and forth until the songs are spaced in a way that is pleasing and exciting. Some songs will lead naturally into the next, others will want a period of silence so that there is a real dramatic impact when the song kicks in, and so on. When in doubt, have the song begin on the downbeat of the next "invisible" measure after the last song. Inserting markers can be useful for this. In Sonar, you can hit F11 during playback to insert a marker in real time. Listen to the tail end of song 3, and hit F11 when you think song 4 should start. then drag song 4 to that spot. Concurrently with this, you should be adjusting the fade ins/fade outs of your songs to suit the tempo and feel of the song, but also the tempo and feel of the album. Unless one song feeds right into another (as in a crossfade or a live set with audience noise or whatever), almost every song should start and end with a fade. It might be a very fast fade, but a fade will prevent clips, pops, or jarring transitions in ambient sound from one track to the next. As a rule of thumb, fade outs usually start slow and then speed up, and fade ins are just the opposite.

Song order:
Listen to your songs in order and second-guess yourself. Listen especially for jarring or unbecoming transitions between songs. You definitely want to have your very best song at or near the beginning of the album if you want anyone to bother listening to the rest of it. (again-family and friends might not tell you how good you really are, but they will usually give pretty consistent responses if you ask them which songs are better/worse than others). The album, regardless of genre, should usually start with the catchiest, most accessible song, the one that's easiest to get into. An exception might be if there is one outright pop song on an album of music that is not otherwise a "pop" record-- as much as you want people to listen past the first few seconds, you also don't want to piss off or alienate the people who might become your biggest fans. It's also not a bad a idea to put a song at the end that everyone seems to like-- makes people want to hit repeat. For the stuff in-between, try and set up the song order as though it were "sets" in a live show.

So now you've got all your songs in order, they sound good in order, they sound good together on shuffle, they are balanced and proportionate, and everything is happy and hunky-dory at 83dB SPL. Time to do the technical stuff.

Home Mastering Stage 2- adjusting overall level. From here on, you are ONLY allowed to make changes to THE WHOLE ALBUM, not to individual tracks. You may only adjust individual tracks in relation to EACH OTHER. If you need to change a song, you go back to stage 1 and start over. Got it? Good.

Overall level:
Now that the individual songs are set in levels that are proportionate to each other, play the whole album through and watch the meters. Figure out which song has the loudest average RMS level. (you can also use some kind of analyzer tool for this). It may be that there is a particularly loud section of an otherwise quieter song that you need to go by. You want to basically figure out where the “loudest part” of the album is. This will always be somewhat subjective. Make sure that you are going by RMS level, and NOT peak level. While you’re at it, identify the quietest song on the album, RMS-wise.

Once you have identified the loudest overall section (by RMS), turn up the WHOLE ALBUM so that THAT SONG is PEAKING at about -0.3 dB. Don’t worry about the levels of anything else just yet, just adjust the volume of the whole project so that the loudest song, RMS-wise, is coming in just under a digital over, PEAK-wise. For the moment, we ONLY care about the LOUDEST part of the album, RMS-wise. Make sense? This is important.

Now check the album, all the way through, and watch for digital overs. If everything was well-mixed with controlled, natural, balanced dynamics throughout, then there will be no clipping on the whole album. If so, great, you did awesome, move on to the next step, “Stage 3-playing with others.”

Probably, though, there will be one or two digital overs, here and there. Maybe one of the quieter songs has a loud snare, or maybe there is a section where the electric bass is turned up for a solo and pins the meter. Don’t sweat it, yet.

What you want to pay attention to is 1. How LONG the digital overs are, and 2. What the average RMS volume is of the quietest song and the loudest song.

WARNING: We are about to something that can be very dangerous to the quality of your audio. It is something that, overused, can seem to the untrained ear like an instant, across-the-board improvement in sound quality, but that is a psycho-acoustical illusion. We are about to apply limiting. A limiter limits the peak volume of the program material (aka the song). This allows us to turn up the average (RMS) volume before clipping. Program material with a louder RMS volume, in the short term, almost always sounds better than quieter stuff. But if the only reason it is louder is because the dynamic range has been unnaturally limited, then it quickly becomes fatiguing and headache-inducing to listen to. THE INSTANT YOU APPLY SEVERE LIMITING WITH MAKEUP GAIN, THE SOUND QUALITY WILL SEEM TO IMPROVE. BUT THAT IS ONLY BEACAUSE IT IS LOUDER. AFTER MINUTES, THE SOUND BECOMES GRATING AND TURN-OFF-ISH, BECAUSE THE DYNAMICS ARE UNNATURAL AND HARSH. Your audience probably has a volume knob. Let them use it for what it was intended for. Do not use limiting to make your material seem louder than it is.

Ideally, in this method, you want to end up with the average level of the quietest song coming in not much lower than -20dB RMS. You are absolutely forbidden to make any corrections that make your loudest song louder than -12dB, RMS. If the quietest section of the quietest song averages -24 RMS, and the loudest section of the loudest song averages -14 RMS, and the peaks are coming in at -0.3 with only near-instantaneous limiting, then you probably have an excellent, well-balanced, dynamic mix. If any song has less than 12dB dynamic range between the average and peak levels, then it is probably going to give your audience a headache if they listen to it at normal volume.

This requires careful judgment on your part. A limiter is the audio equivalent of a loaded gun. It is what makes the world safe, and also what makes it dangerous. It is what makes most home-mastered records sound like dog excrement. But it is often a necessary evil if you have a record that sounds great, but has one snare hit that is 6dB louder than everything else.

Remember above when I said you want to keep track of how long the overs are? You are only allowed to use the limiter only on overs that are less than 2ms long, and that long only in extreme circumstances. The ideal scenario is that your overs are only occasional instantaneous transients lasting just a sample or two (in which case, my advice is to just leave the clipping in there-- that’s right, skip the limiter and let it clip for one or two samples). If you can bring your level up further and have only two- or three-sample clipping on occasion, then go a ahead and do it. Use the limiter if you want, but 99.9% of A/D converters in the world will gloss over digital clipping of only a couple samples, and almost none of them will reproduce the dynamic range differences of a one-sample sound, never mind the speakers or the amp.

What you want is to achieve the loudest level you can get without artificially squishing the dynamics. What happens if there are short portions that clip for audible periods, such as a loud bass note in one song? Here and only here, you are allowed to use bandwidth-limited compression. Set up the compressor with a really fast attack and a ratio of 2:1, maybe 3:1 in extreme cases. Set the threshold to about -2dB, maybe -4 if the clipping is bad. Set a fast-ish release time with a narrow bandwidth of maybe 1/3 octave or less. Loop the clipped section and play with the frequency until the peak level is lowest. Then narrow the frequency as much as you can and play with the release time to make the compressor sound most transparent. Bring up the ratio until just before clipping, about -.03dB. Set up automation so the compressor only kicks in on that note.
What if, on the other hand, there are lengthy portions that clip repeatedly? You have to remix. Sorry, rules are rules.

Again, do yourself a favor and keep your dynamics control to the absolute minimum required to get the loud sections loud. You can limit as many 3-sample overs as you want, but you are only allowed two instances of limiting for more than 1ms per song, and none for more than 3ms. Bandwidth-limited compression is permitted ONLY for individual notes at this stage.

Home Mastering Stage 3- playing with others.

In this stage, we will compare our master against commercially-mastered CDs to make sure we’re not kidding ourselves. Again, in this stage, you are not permitted to correct individual songs. The reasoning will be explained later. Any changes that you make have to be made to the WHOLE ALBUM.

Now that your album sounds well-balanced and well-put-together and is up to a reasonable listening level with a few well-controlled overs, it is time to pick a CD, any CD-- well, not just any one. Pick one that is similar to the style of music that you are doing, and that “sounds” the way you want yours to sound. Pick two, if you want. In fact, it is even better to pick two or more that have a similar vibe and dynamic to what you want yours to have, and it’s not a bad a idea to pick a third that you think most of your target audience will be listening to. Rip those Cds to stereo .wav files and add them to your project as tracks two and three and whatever.

Now, before you hit play, remember what we said about the loudness effect back in step two. U2 and Guns N’ Roses and Nelly and Madonna all have major advantages that you don’t have, and their mixing and mastering engineers have likely squeezed out a few extra decibels more than you will be able to without compromising audio quality. Like an amateur playing against professional golfer, you deserve a handicap. Not only is this fair, but having the handicap will probably improve your level of play by increasing your comfort level. So drop all those “pro” Cds by 6dB before you compare them to your own. I personally guarantee, nobody will ever not buy your record because they had to turn up their stereo by 6dB to hear it.

Now play back the “reference” material and compare it to your own. Does yours sound well-balanced? Do the loud songs sound loud, and the delicate ones delicate? Is the bass powerful and clear? Are the highs articulate and smooth? Is the midrange (most important) clean and well-defined? Are the vocals clear and present? Are the drums impactful and appropriate to the mix?

If you want, at this stage, you can make additional broadband, low-level eq corrections, using broad, shallow, low-Q cuts/boosts. Here, you are allowed to use filters of up to +/-6db, but only ones that cover at least two octaves. You can make small (up to 3dB), subtle corrections to the limiter or compression settings that you applied above, but only to correct transient overs, not to squash the performance dynamics. If particular songs need work, then you MUST fix them in the mix, or in the stage 1 processing (comparing them to other songs in the album, not to outside masters).


End notes:

The reasoning behind all the rules I have laid out is to protect you from doing more harm than good. As I said at the beginning, the primary job of the mastering engineer is to fix the stuff that mixing engineer cannot control. There is no such thing as a good mixing environment that is also a good mastering environment. 9 times out of ten, the final mixes sound better than home-mastered stuff does. The listener may have to adjust their volume knob or tone controls more often, but who cares? If you had to ask about mastering, then I guarantee that your room and/or equipment and/or ears are not up to the task.

I am not trying to insult you (the same is true for me), I am merely trying to point out that, if your room has a -10dB cancellation at 200Hz (and a 10dB cancellation is not at all uncommon, even in pro studios with the best gear), then you will probably compensate for that at the recording stage, at the mixing stage, and at the mastering stage, making things progressively worse at every stage for any listener who is not sitting in your room with his head between your speakers. In reality, you probably have much worse problems than a single 10dB room cancellation at one frequency, and to be honest, your gear and your ability are probably first on the list. Mixing is a mostly aesthetic challenge-- you hear the music, you mix it to get the right sound. But mastering is a more technical, more refined art. The job of the mix engineer is to make it sound great RIGHT NOW. The job of the mastering engineer is to make it sound the way it sounded to the mix engineer EVERYWHERE AND ALL THE TIME. This is the difference between photography and a photo-processing lab.

You will hear people tell you to master every song to a certain level, or to use this effect or that effect for best results. This type of one-size-fits-all approach is wrong.

The best thing you can do is to mix all your songs to sound as best you can, and then limit your “home mastering” to making them balance out with each other and sound good compared with commercial CDs. Commercial Mastering can be surprisingly inexpensive, and it will usually yeild much better results.

Cheers.

<message edited by yep on May 25, 05 10:55 PM>

Address the process rather than the outcome.  Then, the outcome becomes more likely.   - Fripp

Elantric


Bill Ruppert

He should have worked on his voice over first!
The pops and syllabus (s-sounds) sounds on his voice are bad.



Quote from: Elantric on November 01, 2013, 09:24:05 AM
Many good educational tutorials on EQ and Mastering  here
https://www.youtube.com/user/Dr77Music?feature=watch



Brak(E)man

swimming with a hole in my body

I play Country music too, I'm just not sure which country it's from...

"The only thing worse than a guitar is a guitarist!"
- Lydia Lunch

Toby Krebs

Having spent much of the last year recording electric guitars on an album recorded and mastered in a medium sized studio and being involved in a couple of mixing/ mastering sessions I can say that it is a big pain in the ass to get somewhat right. Watching the songwriters agonize and argue with the producer and engineer and other musicians was tough although handled in a very healthy way.Having done 5 or 6 albums and tons of commercials in various studios I  have learned that a good part of mastering is not ruining what has been recorded. If it was recorded well in the first place it will usually turn out o.k. Mastering should always be considered throughout the entire recording process. IMHO a mastering engineer does only what is needed and should not put his "stink" on to the artists record. I have seen and heard entire albums ruined because of this happening. Not good.



Autana

Quote from: Toby Krebs on November 04, 2013, 08:35:23 AM
.Having done 5 or 6 albums and tons of commercials in various studios I  have learned that a good part of mastering is not ruining what has been recorded. If it was recorded well in the first place it will usually turn out o.k. Mastering should always be considered throughout the entire recording process. IMHO a mastering engineer does only what is needed and should not put his "stink" on to the artists record. I have seen and heard entire albums ruined because of this happening. Not good.

Agree Toby! that is, many years ago I read about the concept of to "preserve" rather than "transfigure" which drive sometimes to a embarrasing "mutant" , one of the sources was The Mastering Engineer's Handbook by Bobby Owsinski






GR-55, GP-10, GI-20, Godin xtSA, GodinNylon MultiAc, Giannini classical, 3 GK-3'd gtrs, Cube 80XL, Primova GKFX-21 (x2)

Fear just pulls you out of being true to music, which is coming from a place of love. Love is the opposite of fear. I stay away from anything fear-related.
- Tal Wilkenfeld -

chrish

I've been reading alot lately about this and doing alot of trail and error manipulations to my current ambient recording project within both audacity and reapear. I had about 4 clipping peaks that occured while recording that were somewhat fixed by using a limiter set at a theshold of -4 db to the entire song, which runs about 31 minutes. Before doing this, i cut a shallow eq bell curve of a max of -4 db, from 150 to 300 hz. I read that this is where mud builds up in the reverb tails and it certainly improved the high frequency definition (without having to boost the highs) of my project. The project sounds good through headphones but very poor on a stock car system, but this is simular to the way other commercially produced ambient recordings sound to my ear.  My room is poor and i don't have reference monitors so i've been doing this 'home mastering' with headphones and wondering how my project will translate. Also wondering  how much difference using a 6 thousand dollar eq (and other expensive mastering hardware) vs using the software eq's makes. I suspect alot. What's a poor boy to do?

thebrushwithin

Without spending any money, there is a fairly easy way to come up with acceptable mixes. Use music that you would like to emulate. for example, if you consider your song to be in the same genre as Vangelis, and there is a particular song that you wish your song to sound like, just use the eq, and other sound shaping tools to have your song sonically, like the Vangelis piece. Just listen to the Vangelis piece through the same monitors, until your piece has the same sonic characteristics.

chrish

Brush Within. Your responce has me wondering if there is a  tool where a person could model the eq and other sound shapes of a particular tune, like a Vangelis tune, and then  apply those settings to any other tune? Kind of like how the kempler amp does it's modeling. I did come across a reference in another internet site for a mastering service that employs a computer program to do the mastering. i think it's called 'Landr', if my memory is correct.

Elantric

Quotehas me wondering if there is a  tool where a person could model the eq and other sound shapes of a particular tune, like a Vangelis tune, and then  apply those settings to any other tune? Kind of like how the Kemper amp does it's modeling. I did come across a reference in another internet site for a mastering service that employs a computer program to do the mastering. i think it's called 'Landr', if my memory is correct.


https://www.landr.com/en?
QuoteHow can I prepare my mix to get the best results from mastering?

LANDR is designed to be nearly foolproof, but, like any mastering service, or relationship, benefits from a little bit of breathing room.

First of all, make sure the master output isn't clipping. In fact, you should leave several dbs of headroom (keep the bouncing master level meters well below the red, ideally at -4 or -5dBs).

Secondly, avoid compression, limiting or EQ on the master output of your final mix. Only apply these processes if you feel they are musically essential, and try a version where they are left out completely.

Finally, export your mix at the highest possible bit depth (ex: 24bit) and sample rate (ex: 48kHz). This shouldn't make a huge difference over 16bit / 44.1kHz, but the more resolution LANDR has to read from, the more precisely it will make its adjustments.

Can a robot really mix or master my tracks?

LANDR is a tool designed by sound engineers, so although it employs artificial intelligence, it is guided by profoundly human thought. The LANDR team is comprised of leading sound engineers, signal processing specialists, record label owners, musicians and even an astrophysicist, all pursuing excellence in sound using cutting edge-technology.

I don't like the idea of my track being over compressed and slammed through a limiter to sound loud at all costs. Is that what LANDR does?

Not at all. We've worked hard to avoid this kind of simplistic mastering and have created a highly responsive, sensitive and incredibly musical mastering engine. If you think that our masters still sound too loud, try adjusting the settings. Keep in mind that we've tuned the output based on real human testing, so LANDR is designed to give users what they want... and it's listening and learning to all the feedback we throw at it, so it's only going to get better.

Does it only master songs?

Nope, in fact we encourage people to try different things on it. Master samples for production work, drop in a drum bus, process a vocal on its own, or even put a full DJ set or live performance in. Get creative with LANDR and let us know how it sounds!

Should I use LANDR, or hire a mastering studio?

The polish and balance achieved through the subtle adjustments of a skilled mastering engineer is not something we would ever diminish. There are some very talented engineers out there with exceptional ears, capable of brilliant work. It's a beautiful art form, but there's also a lot of science behind it, as our resident astrophysicist can attest. What we've designed is a tool that's so easy and affordable to use, it just makes sense to try LANDR out first. If you're unsure of the results, and can afford to invest more time, effort and money, then pro mastering may well be worth your while. Having a LANDR file as a reference can be a useful tool; along with your feedback it will guide a mastering engineer in the right direction.

Some other options:


HarBal Spectrum Analysis Software - audio tweakers tool use this to nail that tone
http://www.har-bal.com/

How do you use a Frequency Analyzer to help model guitars with the VG-99?
https://www.vguitarforums.com/smf/index.php?topic=2429.msg14053#msg14053

Real Time Audio Analyzer?
https://www.vguitarforums.com/smf/index.php?topic=54.msg22140#msg22140


Brak(E)man

Quote from: chrish on April 19, 2016, 06:45:08 PM
Brush Within. Your responce has me wondering if there is a  tool where a person could model the eq and other sound shapes of a particular tune, like a Vangelis tune, and then  apply those settings to any other tune? Kind of like how the kempler amp does it's modeling. I did come across a reference in another internet site for a mastering service that employs a computer program to do the mastering. i think it's called 'Landr', if my memory is correct.
Quote from: chrish on April 19, 2016, 09:10:52 AM
I've been reading alot lately about this and doing alot of trail and error manipulations to my current ambient recording project within both audacity and reapear. I had about 4 clipping peaks that occured while recording that were somewhat fixed by using a limiter set at a theshold of -4 db to the entire song, which runs about 31 minutes. Before doing this, i cut a shallow eq bell curve of a max of -4 db, from 150 to 300 hz. I read that this is where mud builds up in the reverb tails and it certainly improved the high frequency definition (without having to boost the highs) of my project. The project sounds good through headphones but very poor on a stock car system, but this is simular to the way other commercially produced ambient recordings sound to my ear.  My room is poor and i don't have reference monitors so i've been doing this 'home mastering' with headphones and wondering how my project will translate. Also wondering  how much difference using a 6 thousand dollar eq (and other expensive mastering hardware) vs using the software eq's makes. I suspect alot. What's a poor boy to do?


I would recommend a person that have experience over any expensive equipment or mechanical tools.

Mastering is today about 50% of the sound if done correct.
So far nothing has come close to experience.
Forget about 200000€ eq
swimming with a hole in my body

I play Country music too, I'm just not sure which country it's from...

"The only thing worse than a guitar is a guitarist!"
- Lydia Lunch


chrish

I gave landr a test drive today. It gave me a free low quality mp3 master (medium setting) as promised on a 4 minute ambient tune i submitted.

I ran some frequency analysis on several sections of the audio wave forms. My original wav file has very little to zero audio content above 10000 Hz in most sections. The landr master mp3 boosted from 10000 Hz to 12000 Hz by around +3db. It also boosted from 15000 to 18500 Hz from +3 to +5db.

The result is, you guessed it, a very noticable audible hiss. I don't think that a human engineer would  boost audible hiss at a frequency where no audio content exists. 

I then tried to feed landr a large 314 MB wav file, which is one 31 minute song on my ambient project. Landr did the file upload and showed that it had completed the master and then said sorry, try again. It did this twice.

Sorry Landr, i may not try again, but i will suggest to them that they give you a free small sample, like a minute of a higher quality wav file so a potential customer could do a proper analysis, both an A/B listening and waveform.

Edit paragraphs

chrish

Here are two wave snippets.. One is the Landr mp3 master and one is a quick mp3 master that I did. Which one sounds better? Which one is the Landr clip?

gumtown

because they are MP3, not really doing it for me.
but WoP track sounds more refined, the PoM sounds a bit too compressed for me.
I had a go at ruining your clip just with some basic tools,
instead of boosting nothing at the high frequencies 12K+ something is there to cut instead of boosting noise and hiss..
Free "GR-55 FloorBoard" editor software from https://sourceforge.net/projects/grfloorboard/

Brak(E)man

The problem at hand is that a tool can't replace a person , not yet anyway

It simply lacks psycho acoustics
intuition and experience
and that'll take a while before that's
incorporated
swimming with a hole in my body

I play Country music too, I'm just not sure which country it's from...

"The only thing worse than a guitar is a guitarist!"
- Lydia Lunch

chrish

Quote from: gumtown on May 02, 2016, 08:40:50 PM
because they are MP3, not really doing it for me.
but WoP track sounds more refined, the PoM sounds a bit too compressed for me.
I had a go at ruining your clip just with some basic tools,
instead of boosting nothing at the high frequencies 12K+ something is there to cut instead of boosting noise and hiss..
Yes, I agree with you about the mp3 quality. The PoM mp3 was the Landr master, and there is,, to my ear a pronounced increase in the audible hiss. Good ear on hearing the compression from such a small sample. From looking at the waveform in my DAW, it's clear that Landr boosted the over all volume gain which would have clipped with out the comp/limiter.

There is still some hiss on my original (Wop mp3) which I cut -3db in the 160 to 300 Hz range and did a -2 db cut in gain at the very low and very high frequency areas (cut 20hz and 20k hz).

I do like the clarity that you added to the clip. Could I ask what tools you applied?
Quote from: Brak(E)man on May 03, 2016, 12:27:25 AM
The problem at hand is that a tool can't replace a person , not yet anyway

It simply lacks psycho acoustics
intuition and experience
and that'll take a while before that's
incorporated

It appears to me that Landr is still in the beta testing stage and that the company would like input on their masters so they can make adjustments. Kind of like testing computers (AI) against master chess players, Jeopardy games show games and that Chinese board game that is being touted as having more moves then atoms in the universe (unlikely IMO as one is finite and the other is infinite).

So I guess if you join Landr and pay for the service, you are paying them to beta test their product. It's not to expensive at the moment, so I may spring for a higher quality file, that is if it can handle the 314MB wave file that is the main body of my ambient project.


Brak(E)man

Quote from: chrish on May 03, 2016, 09:04:29 AM


So I guess if you join Landr and pay for the service, you are paying them to beta test their product. It's not to expensive at the moment, so I may spring for a higher quality file, that is if it can handle the 314MB wave file that is the main body of my ambient project.

I'm talking several years maybe decades,
Since we don't understand psyco acoustics
nor intuition , or can teach a machine
years of experience of something
that really is a matter of taste apart from the technical aspects
how would the computers or programs of today cope.
You can probably teach a program to recognize non musical hiss
but to recognize musical hiss ....

And I'd cut your file way before 20 Hz or 20 KHz
and remove the hiss with a plugin for noise
Waves for example
swimming with a hole in my body

I play Country music too, I'm just not sure which country it's from...

"The only thing worse than a guitar is a guitarist!"
- Lydia Lunch

chrish

I had sent landr an e-mail about what i posted here and received a reply that said ''for best results, you need to mix into landr.'' They suggested that i lower the db, reduce compression on individual tracks and avoid using limiters. Well the song they ''mastered'' for me was a one track (not a mix of separate recorded tracks), had headroom, zero compression and no limiting and the thing wants to allegedly blame me for the poor result. :-) I think i may have gotten a form letter. As i recall back in 2001, didn't the  HAL 9000 series computer display a simular attitude. :-)

chrish

Here is a link for a more comprehensive review of LANDR.

http://passivepromotion.com/can-landr-replace-your-mastering-engineer.

Quote: "1. Am I biased? Perhaps, but who's more qualified to evaluate a mastering service than a mastering engineer? Let your ears be the judge.

2. Louder pretty much always sounds better to the human ear. While differences in volume are important for the purposes of this evaluation, you should try to match playback levels when comparing the examples for sound quality.

3. This is an apples-to-apples comparison. I paid $9 for four uncompressed 16-bit .wav masters, to compare to my uncompressed 16-bit .wav masters. All files were encoded to mp3 at 320 kbps using LAME at the highest quality setting.

Is it just me, or does the LANDR version sound like it was mastered by an astrophysicist?

To my ears, the biggest problem is the lack of low end. The result is thin and narrow and just "off."

Keep in mind, LANDR uses the same algorithm for all genres of music. This tonal balance might be perfect for folk or classical, but it doesn't cut it for EDM, hip hop, or even pop.....

....On the plus side, I appreciate that LANDR doesn't win the Loudness War by default. There's plenty of dynamic range. Unfortunately, the track would be the quietest in any EDM playlist. It's simply not club-ready...

It's easy to trade low frequency energy for volume. The challenge is achieving both.

Worse, this version is even more compressed (as opposed to peak limited). You can see visually how little dynamic contrast there is, compared to my master at pretty much the same volume."