The state of audio engineering today
I'm actually supposed to be studying for my exam but oh well. I frequent the Gearslutz mastering forum sometimes, and i'm astonished on how much misinformation there is coming from so-called "established proffesionals" - most of which actually do have a studio and offer mastering services.
Me, i can't call myself a mastering engineer since i haven't worked on such jobs at a pro level, but i'm doing recording and mixing every now and then. Sure, it doesn't bring me much money, but at 19 and with just a handful of tools you gotta start somewhere. And since the main point of "mastering" today is just to "make it loud", i'd rather skip it altogether. Who wants that is free to take the recording he made at my place and pay someone else to do it.
Anyway, i see "engineers" that have no idea what RMS levels are and how they are measured! I see talks about stuff being mastered at -4 RMS which is physically impossible. Pure sine wave = -3 RMS, full bandwidth white noise = -4.8 RMS. A RMS measurement by definition is a longer term average, so if the meter you are using has a short window it's obvious that it'll misguide you. AFAIK the standard is the average value for the whole track, and a healthy value (specified by the BBC if i recall correctly) is -12 RMS. Most quality albums sit in the -18 to -12 ballpark. Yet the trend has increased to a -9 average, with occasional -7 and even -6. But let's leave the loudness war alone for now, it's been discussed too many times.
Most stuff that uses overtone-rich synths tends to sound like white noise in the highs. But it doesn't get any louder than that no matter what you try. Eminem's Recovery, a brilliant album in terms of lyrics and flow, sits around -6 RMS and it SOUNDS LIKE ASS. Not only is it midrange shouty (we're already used to seeing that from Em), there's more than obvious distortion and almost zero bass on the whole thing.
But really, this is as loud as it'll go, period. IMO in anything to do with audio, levels and tone, one needs to do two things:
1. USE YOUR EARS.
2. If there's something you're unsure of, USE YOUR EYES. Analyze the waveform both offline (after rendering) and "live". And always use outboard metering if possible, both peak and RMS/VU. The only thing you need a digital meter for is to be sure that you're not clipping. The only digital meters i watch are the attenuation meters of the compressor and limiter, so i know when i'm overdoing it. There are few good articles about compression on the net, but from the ones i found useful i always remember this one: If the compressor gain reduction meter doesn't return to zero after each bar in the song, you're overdoing it. Another rule i go by is if i need to limit more than 3dB it needs to be fixed in the mix. I do however favor soft clipping instead of limiting, so naturally i need to keep a closer eye on it to keep the distortion low. However, i refuse to do anything over -12 RMS, and that keeps my tracks nice and clean.
My mixing console has old-style analog VU meters, and that's what i use as reference. I've grown around gear with VU meters, and what i loved since i was a kid was to turn up the bass knob to make the meters swing full scale and back. And over time i found that the bass knob needed to go higher and higher to achieve this feat - to make anything louder you inevitably have to sacrifice bass response. On a few recent albums, with no EQ applied, the VU meters barely move at all. They just stick at one level from end to end. It's crazy.
It's ironic that we need upgraded speaker systems and lossless audio codecs to listen to crap music. Good old 128kbps MP3 sounds awesome on a song with high dynamic range and sparse composition, because the less material to encode, the better the encoder can conceal the artifacts. And it's counterproductive to work hard and mix lots and lots of stuff in your song - when you slam it afterwards and it all turns into a mess, but people still do it. Worse, digital limiting has brought a new problem - the lowpass filter of the MP3 encoder overshoots and clips. You need at least 0.5dB headroom to ensure that the MP3 ends up clean - yet most digital limiters default to a ceiling of like -0.03 - and i'm not sure such differences would even encode in 16-bit, let alone be audible.
Some stuff sounds good indeed when slammed, but most doesn't. And it gives a headache too. Well, i dunno what you think but i just want my bass back.
Me, i can't call myself a mastering engineer since i haven't worked on such jobs at a pro level, but i'm doing recording and mixing every now and then. Sure, it doesn't bring me much money, but at 19 and with just a handful of tools you gotta start somewhere. And since the main point of "mastering" today is just to "make it loud", i'd rather skip it altogether. Who wants that is free to take the recording he made at my place and pay someone else to do it.
Anyway, i see "engineers" that have no idea what RMS levels are and how they are measured! I see talks about stuff being mastered at -4 RMS which is physically impossible. Pure sine wave = -3 RMS, full bandwidth white noise = -4.8 RMS. A RMS measurement by definition is a longer term average, so if the meter you are using has a short window it's obvious that it'll misguide you. AFAIK the standard is the average value for the whole track, and a healthy value (specified by the BBC if i recall correctly) is -12 RMS. Most quality albums sit in the -18 to -12 ballpark. Yet the trend has increased to a -9 average, with occasional -7 and even -6. But let's leave the loudness war alone for now, it's been discussed too many times.
Most stuff that uses overtone-rich synths tends to sound like white noise in the highs. But it doesn't get any louder than that no matter what you try. Eminem's Recovery, a brilliant album in terms of lyrics and flow, sits around -6 RMS and it SOUNDS LIKE ASS. Not only is it midrange shouty (we're already used to seeing that from Em), there's more than obvious distortion and almost zero bass on the whole thing.
But really, this is as loud as it'll go, period. IMO in anything to do with audio, levels and tone, one needs to do two things:
1. USE YOUR EARS.
2. If there's something you're unsure of, USE YOUR EYES. Analyze the waveform both offline (after rendering) and "live". And always use outboard metering if possible, both peak and RMS/VU. The only thing you need a digital meter for is to be sure that you're not clipping. The only digital meters i watch are the attenuation meters of the compressor and limiter, so i know when i'm overdoing it. There are few good articles about compression on the net, but from the ones i found useful i always remember this one: If the compressor gain reduction meter doesn't return to zero after each bar in the song, you're overdoing it. Another rule i go by is if i need to limit more than 3dB it needs to be fixed in the mix. I do however favor soft clipping instead of limiting, so naturally i need to keep a closer eye on it to keep the distortion low. However, i refuse to do anything over -12 RMS, and that keeps my tracks nice and clean.
My mixing console has old-style analog VU meters, and that's what i use as reference. I've grown around gear with VU meters, and what i loved since i was a kid was to turn up the bass knob to make the meters swing full scale and back. And over time i found that the bass knob needed to go higher and higher to achieve this feat - to make anything louder you inevitably have to sacrifice bass response. On a few recent albums, with no EQ applied, the VU meters barely move at all. They just stick at one level from end to end. It's crazy.
It's ironic that we need upgraded speaker systems and lossless audio codecs to listen to crap music. Good old 128kbps MP3 sounds awesome on a song with high dynamic range and sparse composition, because the less material to encode, the better the encoder can conceal the artifacts. And it's counterproductive to work hard and mix lots and lots of stuff in your song - when you slam it afterwards and it all turns into a mess, but people still do it. Worse, digital limiting has brought a new problem - the lowpass filter of the MP3 encoder overshoots and clips. You need at least 0.5dB headroom to ensure that the MP3 ends up clean - yet most digital limiters default to a ceiling of like -0.03 - and i'm not sure such differences would even encode in 16-bit, let alone be audible.
Some stuff sounds good indeed when slammed, but most doesn't. And it gives a headache too. Well, i dunno what you think but i just want my bass back.
Comment