I found a song with the bass boosted, and it sounds perfect. I wanted to use the method used on the song for other songs, but I do not know the method and everything I have tried is not as good. The song is Lil Wayne’s Krazy, and I was wondering if it was apparent what steps I should take to emulate the bass boosting method used. Bass boosting alone doesn’t work, and using low/high pass filters doesn’t seem to work.
I am running Windows 74, 64 bit, Audacity 2.0.5
The top track is unboosted, the lower is what I’m trying to accomplish.
I think it needs some kind of leveler because the waveform is pretty straight across.
As with the whole plot spectrum, the lower frequencies need to have a higher db and higher frequencies need a lower db.
And of course, bass needs to be increased to a peak of -27.8 db for 50 frequency.
Any help is greatly appreciated. I’m really just looking for a clean way to boost bass that won’t blow my speakers/sound distorted.
The flat-top dense waveforms indicate limiting or perhaps clipping (distortion). Even the top waveform isn’t what “natural music” looks like. The bottom waveform is “worse”. i.e. If you record live music played with real instruments, it’s a lot more dynamic (with louder & quieter parts). I’m not saying it’s bad if that’s what you want. A lot of modern music is produced like this. (See [u]The Loudness War[/u].)
Some background - Most audio formats (as well as your digital-to-analog converters) are limited to 0dB (=1.0 =100%). However, Audacity (like most audio editors) uses floating-point internally, so it can “temporarily” go over 0dB. That means if you just boost the bass, you might not hear the clipping/limiting unless you have the playback volume set to 100%, or until after you save the file as a normal 16-bit or 24-bit WAV.
After boosting the bass, try the Hard Limiter effect set to 0dB. If that sounds too harsh or distorted, you can try the Compressor effect. There are several settings to experiment with, but normally a compressor (and/or limiter) is used to boost the overall volume without clipping/distorting the peaks. I believe Audacity’s Hard Limiter effect is really a hard-clipping effect so you may want to look for a better limiter plug-in that can “round over” the peaks without as much distortion… I think that’s what you really need… A good limiter applied after boosting the bass.
You might also try using a high-pass filter to remove the really deep bass (below 40 or 50 Hz). If you don’t have a big amplifier and big woofer/subwoofer, any deep bass that your system can’t reproduce just creates distortion. And just for reference, the lowest note on a normally-tuned bass guitar is around 40Hz, and big subwoofers used for dance clubs & live performances usually “only” go down to around 40Hz… That’s deep enough for a good thump you can feel! (If you have the appropriate woofer & amp.)
You probably cannot achieve the “Lil Wayne” sound, but you may be able to get close to it. It has to start with the instruments & music on the recording… If another recording has different bass frequency & tone to start with, there is only so much you can do. The recording engineer can also process the bass guitar & kick drum separately before mixing with the vocals & other instruments… You can’t do that. And, the recording engineer & producer has access to high-end plug-ins & tools and the skill & experience to get the desired sound out of it.
I’m really just looking for a clean way to boost bass that won’t blow my speakers…
I can’t guarantee that you won’t blow your speakers! If you start boosting & compressing/limiting the bass, you are increasing the average power to your woofers, so if your amp is capable of blowing your speakers, it could happen.
I listened to LiL Wayne’s Crazy official YouTube video and see it applies a resonator type bass frequency. If you haven’t already found the following YouTube tut linked below on how to increase such bass frequencies follow the instructions to the letter and take note of how he chooses the Audacity Amplifier and Plot Spectrum to apply the right amount of EQ which he applies twice…
It’s not clean bass. I did a quick analysis of the front of the song and yes it has beats at 50, but also at 150 and 300 (attached). It’s “dirty bass” and they probably cooked it up in the studio. If you just boosted everything in that range it would get boomy and hurt your ears. If you made bass notes at 50Hz, it would sound too mushy and could create speaker damage.
Any engineers/audio experts here ever heard about the waveform RMS (light blue middle section in Audacity waveform) having to stay within the RMS max specs of amp and speaker output?
I’ve heard this mentioned in other audio forums where the poster (didn’t know if he was an expert) said clipping isn’t the most important to be concerned about when playing loud frequencies. It seems to explain why I have some loud bass punchy music in aiff files I purchased on commercial, factory pressed CD’s that shows clipping in Audacity but does not distort on my car audio speakers.
Even the RMS is pretty thin but that boxy sounding bass beat sounds like some giant foot is in the trunk of my sedan trying to kick the back seat out. No distortions. It’s a gorgeous sound that if I’ld played it at current loud volumes back in the '70’s on backdash free air 6x9 Craig Road Rated speakers would surely have torn them to bits.
Also Flaming Lips’ “It Overtakes Me” shows clipping with a huge RMS indicator waveform in its aiff off their “At War With The Mystics” CD which won a 2007 Grammy for Best Engineered Album. That bass line sounds incredible and loud off my back dash Polk 6x8’s with a 80hz bass blocker. No distortion in that I don’t hear electrical spark sound which will happen if I play it much louder.
And here’s the Audacity Plot Spectrum sampled between 1:15-1:19min. of the Overtakes Me song.
It seems it should have a similar resonator sound as LiL Wayne’s “Crazy” seeing 40Hz is at -8db but as you can hear from the YouTube video it doesn’t. I guess music must be heard and edited taking into account the spectrum relationship instead of focusing on just one part of the spectrum.
The signal level, whether “peak” level, “root mean square” (“rms” / “RMS”) or any other measure of signal level, has no relation to the specification of speaker output. Any (non-silent) signal can blow a speaker if amplified sufficiently. There is no “direct” relationship between signal level and loudness - put simply, turn the volume up and any non-silent signal becomes louder.
Digital audio signals are just a series of numbers. For “CD audio standard” (Redbook standard), the numbers are 16 bit signed integers (whole numbers between -32768 and +32767), and there are 44100 numbers per second for each channel (one stream of numbers for the left channel and one stream of numbers for the right channel). These numbers are converted to voltage by the digital-to-analog converter (DAC) which is then amplified to a high current signal by the amplifier. The “numbers” (the digital signal) cannot by itself blow amplifiers or speakers. Too much voltage from the DAC into the amplifier input will blow the amplifier. Too much voltage from the amplifier will blow the speakers.
Audio systems are often designed with amplifiers that are powerful enough to blow the speakers, so that the speakers can be driven close to their maximum rating with a “clean” (low distortion) signal. For such systems, it is the users responsibility to adjust the volume control to a sensible level to avoid damage to speakers.
“Clipped” signals (where the tops/bottoms of the waveform are cut off because they hit the maximum possible level) create high harmonics. It is common for excessive amounts of high harmonics to blow tweeters (high frequency drivers). Clipping distortion not only sounds “bad”, but can damage audio equipment - including ears!
This subject got me to do some more research to confirm what I thought was an industry established standard (in order to prevent damage to audio equipment) of a max peak level of -12db for commercial CD releases which I found your wiki link Redbook standard didn’t make clear. It appears from this article…
…you can go as high as -0.2 dbFS just under clipping. But as you’ve indicated this isn’t what damages audio equipment but the clipping that introduces added harmonics that changes the timbre of the initial sound shape (such as in the form of harsh rumbling in low bass signals that wasn’t there before) that does the damage. So I surmise to keep it safe we should be listening for changes in the character/timbre of the original sound referred to as harmonic distortion caused by clipping during editing.
But this doesn’t explain why the severely clipped first sample I posted doesn’t cause damage or audible distortion on my car’s audio system? Mind you, because it’s already so loud, I only play this at half volume on my car’s 1997 Pioneer CD head unit (from Walmart), through a 250watt Alpine amp (35 watts RMS to each speaker) which is loud enough to be heard over road roar driving 60mph down the highway with the windows down which according to Crutchfield is around 80db. However, any frequencies around 40Hz can’t be heard.
Maybe it isn’t about clipped digital signals at all but about distorting the sound initially with edits, clipped or not, that might cause excessively unbalanced speaker excursion. This would explain why high pitched thonky, thin sounding kick bass sounds from '70’s & '80’s recordings with a -17dbFS RMS limit (according to the article) can’t be amplified on my system without distortion as loudly as today’s big, fat bass signals even when their digital waveform is clipped.
The article says no such things.
Let’s take the extreme example of a clipped signal: A square wave (-1 -1 -1 1 1 1 -1 -1 -1…, at full scale)
The big problem is the fast change from -100 % to 100 % which can cause the speake’s membrane to “overshoot”–always depending on the actual reaction times of the whole system.
The -0.1 to -0.3 dB security head room for inter-sample overs is not even nearly enough (e.g. see the “True Peak” level measurement in the ebu128 recommendation).
You can test it approximately as follows:
Generate a square wave tone (just use the default settings for the other parameters)
read the dB (amplify effect), should be (-)1.9 or so.
call the nyquist prompt and enter
(resample s *sound-srate*)
read the dB peak again.
do the same with a square wave that has no aliasing.
The difference is remarkable.
You can of course test other audio with this trick, however, it isn’t an official method, it only shows what could happen to the output.
I’m fairly convinced that the intersample difference has the greatest impact, apart from e.g. DC offset.
That’s an interesting test, Robert, even though it’s over my head and skill level. IMO music isn’t made and checked for corruption by distortion based on testing one square wave pattern. No one listens or cares what happens to square wave patterns pushed beyond their digitally allotted headroom.
Other than that how do you explain the lack of distortion in the heavily clipped sample I posted? Not here to debate what an article says. This thread is about how to boost bass and make it loud digitally and pinpoint causality to any artifacts as a result to distinguish whether it’s sourced from the file or through system output or a combination of both.
The article just confirms there are commercially released CD’s whose mastering pushed the headroom to -0.2db with no noticeable distortion. I listened to the samples in that article and couldn’t hear any improvement or distortion, they’re just a bit louder and brighter. What’s misleading about that article as is in all of these types of Loudness War discussions is they attempt to prove their premise on songs that are already loud, noisy and peppered with “artful” distortion making it impossible to hear any artifacts caused by brick walling. It’s pointless to make arguments on something no one can hear.
Did those songs need to be brick walled? No. They were loud and noisy to begin with. Does brick walling produce distortion? Well if I can’t here distortion in an intentionally clipped (beyond brick walling) entire waveform on headphones as I posted above and no one can explain why, then I don’t see the point to the question.
I’ll give you an example why testing a square wave isn’t going to hold water musically speaking. I discovered something interesting editing one bass signal wave when applying the Low Pass bass boost method on a duped stereo channel converted to mono. After normalizing the mono bass signal to 0db using Amplify I got a clipped waveform. I zoomed in and selected this wave and used Amplify to remove the clip. The waveform retained its original rounded top (no flat top). I muted the stereo channel and played only the mono bass channel and I got a slight but very audible crackly noise on the repaired waveform. When I applied the same Amplify fix to the same bass wave in the stereo channel there was no crackly sound.