Audio Settings for Recording Voice/Commentary (Questions)

The Majestic

New Member
To any audiophiles out there, I have some questions for you. I am new to the audio aspect of recording and would like some assistance. Here are my specs:

Blue Snowball Microphone
Pop filter
Scissor Arm
Audacity

What I'm trying to do:
1. Trying to find the mic volume best for recording (IE: 0-100 in Windows and Audacity)
2. Prevent Clipping (When I laugh it clips)
3. Understand the following terms: Compression, Limiting, Normalization terms such as those mean, how to use them, in what order to use them and the settings I should use them in.

My Problems:
1. I clip something awful in my recordings.
2. I cannot find out what settings to use Windows and Audacity.
3. I do not know what order to use compression, limiting or normalization, IF I need to even use them.
4. How far should you sit from the mic? I can sit 1 foot from it, or as far as 6 feet.
5. The background noise removal tool, even when used properly results in echoing. How do I fix this?

I know this is a laundry list of problems and they aren't any of yours, however I would greatly appreciate it someone could answer some or all of them for me. I have been doing a lot of testing and research, but I just can't seem to get it right.

Another thing I was wondering is if I needed a better microphone and a mixer/preamp set up.

I like to watch ChilledChaos and Seananners who tend to yell or scream into the mic on occasion. I do not hear distortion or clipping on their videos and I am puzzled at how they manage this. My guess was that the preamp they use has limiting or compression built in, so no sound reaches above -1db, and so it doesn't clip. Would this be a correct assumption?

Does anybody have a setup that they could recommend. I have tried to reach out to several youtubers, however I have received no responses. I assume they are too busy.

TL;DR I would like to be able to speak clearly and concisely, loud enough to be heard over gameplay and quiet enough to prevent clipping during moments where I yell or laugh.
 
D

DJStitch91

Guest
To any audiophiles out there, I have some questions for you. I am new to the audio aspect of recording and would like some assistance. Here are my specs:

Blue Snowball Microphone
Pop filter
Scissor Arm
Audacity

What I'm trying to do:
1. Trying to find the mic volume best for recording (IE: 0-100 in Windows and Audacity)
2. Prevent Clipping (When I laugh it clips)
3. Understand the following terms: Compression, Limiting, Normalization terms such as those mean, how to use them, in what order to use them and the settings I should use them in.

My Problems:
1. I clip something awful in my recordings.
2. I cannot find out what settings to use Windows and Audacity.
3. I do not know what order to use compression, limiting or normalization, IF I need to even use them.
4. How far should you sit from the mic? I can sit 1 foot from it, or as far as 6 feet.
5. The background noise removal tool, even when used properly results in echoing. How do I fix this?

Another thing I was wondering is if I needed a better microphone and a mixer/preamp set up.

I like to watch ChilledChaos and Seananners who tend to yell or scream into the mic on occasion. I do not hear distortion or clipping on their videos and I am puzzled at how they manage this. My guess was that the preamp they use has limiting or compression built in, so no sound reaches above -1db, and so it doesn't clip. Would this be a correct assumption?

Does anybody have a setup that they could recommend. I have tried to reach out to several youtubers, however I have received no responses. I assume they are too busy.

TL;DR I would like to be able to speak clearly and concisely, loud enough to be heard over gameplay and quiet enough to prevent clipping during moments where I yell or laugh.
holy shit that's a list....

**breaths**

I run an at2020 with scarlet 2i2 just so you know... it took me a month-ish to sort my levels out to how i like them.

A1/B2. This you will have to test yourself also try it with auto gain off if it is enabled, as this will always boost your audio too much in my opinion. Does snowball have gain adjustment? if so use it.

A2/B1. Settings are trial and error go learn/discover. Sounds harsh, is the truth, no other way to put it.

A3. Compression: The Compressor effect reduces the dynamic range of audio. One of the main purposes of reducing dynamic range is to permit the audio to be amplified further (without clipping) than would be otherwise possible

Limiting: Use the Limiter effect to pass signals below a specified input level unaffected or gently reduced, while preventing the peaks of stronger signals from exceeding this threshold. Mastering engineers often use limiting combined with make-up gain to increase the perceived loudness of an audio recording during the audio mastering process.

Normalization: Use the Normalize effect to set the peak amplitude of a single track, make multiple tracks have the same peak amplitude and equalize the balance of left and right channels of stereo tracks.

Optionally you can remove any DC offset from the tracks.

B3 What I do is at the end.

B4. Depends on multiple things: if mic is omnidirectional or cardioid. (Solo i recommend Cardiod) and what your levels are at please refer to A2/B1

B5. If its creating echo the settings need tweaking, you could be trying to subtract it in one rather than runnign it multiple times which is lees harsh on the audio. MY SETTINGS: Noise reduction (dB) 12, Sensitivity 6,Frequency smoothing (bands) 0.
OR the echo could be there you just cant recognize it because of the background noise.

I have been doing a lot of testing and research, but I just can't seem to get it right.
all the terms were lifted form audacity's website, have a look above how long it took me to sort out my settings, LITTLE ADJUSTMENTS WORK BETTER. change, test, change, test. not just a couple of mins test at least 30 mins in a game playing with someone not thinking about it, the more you think about it the more you will adjust yourself trying to make it work, which you wont work, you want it to work all the time not just when you are a certain way.

Another thing I was wondering is if I needed a better microphone and a mixer/preamp set up.
No offense but if you can't that mic and settings work a mixer and preamp will make no difference because you will be jumping levels and learning curves. I'd say around 40-60% of amateur/pro tubers use blue mics... use that for your own conclusion.

I like to watch ChilledChaos and Seananners who tend to yell or scream into the mic on occasion. I do not hear distortion or clipping on their videos and I am puzzled at how they manage this. My guess was that the preamp they use has limiting or compression built in, so no sound reaches above -1db, and so it doesn't clip. Would this be a correct assumption?
Their FINAL PRODUCT doesn't clip or distort. you don't know what happens in editing. you can do the same once you've learned how. bottom line you are aiming at BETTER GEAR DOES NOT EQUAL BETTER FINAL PRODUCT, it does however mean it should sound better when USED CORRECTLY.



Ok set up!

AT2020 in homemade shock mount (cause i feel like being ghetto), running through decent wires (cant remember brand) to scarlet 2i2 ran into usb for audacity when solo or ts when not.

2i2 gain is approx 3/4, I run a monitor line into my astro mixamp (which is for my outputs that does not affect mic lines) for when i wont to test mic i fine at levels, halos help a little on the 2i2 but they don't tell the whole story.

On my pc the 2i2 is set too 100% volume with AGC off.

My audio editing process:

In SVP13 i cut all the clips in place to video the record just the mic audio (with extra seconds of no speaking for noise removal.)

Import/open in audacity. find the seconds of no voice, highlight, noise reduction, get noise profile, select all, noise reduction, (enter my settings if you want to test), OK, listen through, if you find piece of backgroud noise, select that section, noise reduction, OK, listen again, push as far as to not affect audio but remember approximately the bottom 10% of volume will be covered by game noise at the same level.

Then Normalize, tick remove dc offset and normalize maximum amptitude to -1.0dB.

Then Compress, Threshold -12dB, Noise floor -40dB, Ratio 2:1, Attack time 0.20 seconds, Release time 1.0 secs. The bottom 2 tick boxes are trial and error which i swap around in combination if it clips or (RARELY EVER) distortion, i 99% run with just compress based on peaks ticked. now compress it multiple times until it is at a reasonable level, good way to check is the drop down arrow on tack and change view to waveform (dB) but default view is fine. Listen and look at your graph after every compress.

Sometimes its ready to roll then but sometimes ill norm to see if it sounds if i don't like it undo re comp then norm again.

Export, back into SVP, replace the audio track there with the new one then list with both game audio and voice over if struggle to hear vo the drop game audio a few dB (should take a lot).


DAMN I WISH THIS SITE GAVE KARMA CAUSE I THINK I DESERVE +1 FOR ALL THIS?!
 

KOB_YT

Member
holy shit that's a list....

**breaths**

I run an at2020 with scarlet 2i2 just so you know... it took me a month-ish to sort my levels out to how i like them.

A1/B2. This you will have to test yourself also try it with auto gain off if it is enabled, as this will always boost your audio too much in my opinion. Does snowball have gain adjustment? if so use it.

A2/B1. Settings are trial and error go learn/discover. Sounds harsh, is the truth, no other way to put it.

A3. Compression: The Compressor effect reduces the dynamic range of audio. One of the main purposes of reducing dynamic range is to permit the audio to be amplified further (without clipping) than would be otherwise possible

Limiting: Use the Limiter effect to pass signals below a specified input level unaffected or gently reduced, while preventing the peaks of stronger signals from exceeding this threshold. Mastering engineers often use limiting combined with make-up gain to increase the perceived loudness of an audio recording during the audio mastering process.

Normalization: Use the Normalize effect to set the peak amplitude of a single track, make multiple tracks have the same peak amplitude and equalize the balance of left and right channels of stereo tracks.

Optionally you can remove any DC offset from the tracks.

B3 What I do is at the end.

B4. Depends on multiple things: if mic is omnidirectional or cardioid. (Solo i recommend Cardiod) and what your levels are at please refer to A2/B1

B5. If its creating echo the settings need tweaking, you could be trying to subtract it in one rather than runnign it multiple times which is lees harsh on the audio. MY SETTINGS: Noise reduction (dB) 12, Sensitivity 6,Frequency smoothing (bands) 0.
OR the echo could be there you just cant recognize it because of the background noise.

all the terms were lifted form audacity's website, have a look above how long it took me to sort out my settings, LITTLE ADJUSTMENTS WORK BETTER. change, test, change, test. not just a couple of mins test at least 30 mins in a game playing with someone not thinking about it, the more you think about it the more you will adjust yourself trying to make it work, which you wont work, you want it to work all the time not just when you are a certain way.


No offense but if you can't that mic and settings work a mixer and preamp will make no difference because you will be jumping levels and learning curves. I'd say around 40-60% of amateur/pro tubers use blue mics... use that for your own conclusion.


Their FINAL PRODUCT doesn't clip or distort. you don't know what happens in editing. you can do the same once you've learned how. bottom line you are aiming at BETTER GEAR DOES NOT EQUAL BETTER FINAL PRODUCT, it does however mean it should sound better when USED CORRECTLY.



Ok set up!

AT2020 in homemade shock mount (cause i feel like being ghetto), running through decent wires (cant remember brand) to scarlet 2i2 ran into usb for audacity when solo or ts when not.

2i2 gain is approx 3/4, I run a monitor line into my astro mixamp (which is for my outputs that does not affect mic lines) for when i wont to test mic i fine at levels, halos help a little on the 2i2 but they don't tell the whole story.

On my pc the 2i2 is set too 100% volume with AGC off.

My audio editing process:

In SVP13 i cut all the clips in place to video the record just the mic audio (with extra seconds of no speaking for noise removal.)

Import/open in audacity. find the seconds of no voice, highlight, noise reduction, get noise profile, select all, noise reduction, (enter my settings if you want to test), OK, listen through, if you find piece of backgroud noise, select that section, noise reduction, OK, listen again, push as far as to not affect audio but remember approximately the bottom 10% of volume will be covered by game noise at the same level.

Then Normalize, tick remove dc offset and normalize maximum amptitude to -1.0dB.

Then Compress, Threshold -12dB, Noise floor -40dB, Ratio 2:1, Attack time 0.20 seconds, Release time 1.0 secs. The bottom 2 tick boxes are trial and error which i swap around in combination if it clips or (RARELY EVER) distortion, i 99% run with just compress based on peaks ticked. now compress it multiple times until it is at a reasonable level, good way to check is the drop down arrow on tack and change view to waveform (dB) but default view is fine. Listen and look at your graph after every compress.

Sometimes its ready to roll then but sometimes ill norm to see if it sounds if i don't like it undo re comp then norm again.

Export, back into SVP, replace the audio track there with the new one then list with both game audio and voice over if struggle to hear vo the drop game audio a few dB (should take a lot).


DAMN I WISH THIS SITE GAVE KARMA CAUSE I THINK I DESERVE +1 FOR ALL THIS?!
You my friend have too much time on your hands xD
 

The Majestic

New Member
holy shit that's a list....

**breaths**

I run an at2020 with scarlet 2i2 just so you know... it took me a month-ish to sort my levels out to how i like them.

A1/B2. This you will have to test yourself also try it with auto gain off if it is enabled, as this will always boost your audio too much in my opinion. Does snowball have gain adjustment? if so use it.

A2/B1. Settings are trial and error go learn/discover. Sounds harsh, is the truth, no other way to put it.

A3. Compression: The Compressor effect reduces the dynamic range of audio. One of the main purposes of reducing dynamic range is to permit the audio to be amplified further (without clipping) than would be otherwise possible

Limiting: Use the Limiter effect to pass signals below a specified input level unaffected or gently reduced, while preventing the peaks of stronger signals from exceeding this threshold. Mastering engineers often use limiting combined with make-up gain to increase the perceived loudness of an audio recording during the audio mastering process.

Normalization: Use the Normalize effect to set the peak amplitude of a single track, make multiple tracks have the same peak amplitude and equalize the balance of left and right channels of stereo tracks.

Optionally you can remove any DC offset from the tracks.

B3 What I do is at the end.

B4. Depends on multiple things: if mic is omnidirectional or cardioid. (Solo i recommend Cardiod) and what your levels are at please refer to A2/B1

B5. If its creating echo the settings need tweaking, you could be trying to subtract it in one rather than runnign it multiple times which is lees harsh on the audio. MY SETTINGS: Noise reduction (dB) 12, Sensitivity 6,Frequency smoothing (bands) 0.
OR the echo could be there you just cant recognize it because of the background noise.

all the terms were lifted form audacity's website, have a look above how long it took me to sort out my settings, LITTLE ADJUSTMENTS WORK BETTER. change, test, change, test. not just a couple of mins test at least 30 mins in a game playing with someone not thinking about it, the more you think about it the more you will adjust yourself trying to make it work, which you wont work, you want it to work all the time not just when you are a certain way.


No offense but if you can't that mic and settings work a mixer and preamp will make no difference because you will be jumping levels and learning curves. I'd say around 40-60% of amateur/pro tubers use blue mics... use that for your own conclusion.


Their FINAL PRODUCT doesn't clip or distort. you don't know what happens in editing. you can do the same once you've learned how. bottom line you are aiming at BETTER GEAR DOES NOT EQUAL BETTER FINAL PRODUCT, it does however mean it should sound better when USED CORRECTLY.



Ok set up!

AT2020 in homemade shock mount (cause i feel like being ghetto), running through decent wires (cant remember brand) to scarlet 2i2 ran into usb for audacity when solo or ts when not.

2i2 gain is approx 3/4, I run a monitor line into my astro mixamp (which is for my outputs that does not affect mic lines) for when i wont to test mic i fine at levels, halos help a little on the 2i2 but they don't tell the whole story.

On my pc the 2i2 is set too 100% volume with AGC off.

My audio editing process:

In SVP13 i cut all the clips in place to video the record just the mic audio (with extra seconds of no speaking for noise removal.)

Import/open in audacity. find the seconds of no voice, highlight, noise reduction, get noise profile, select all, noise reduction, (enter my settings if you want to test), OK, listen through, if you find piece of backgroud noise, select that section, noise reduction, OK, listen again, push as far as to not affect audio but remember approximately the bottom 10% of volume will be covered by game noise at the same level.

Then Normalize, tick remove dc offset and normalize maximum amptitude to -1.0dB.

Then Compress, Threshold -12dB, Noise floor -40dB, Ratio 2:1, Attack time 0.20 seconds, Release time 1.0 secs. The bottom 2 tick boxes are trial and error which i swap around in combination if it clips or (RARELY EVER) distortion, i 99% run with just compress based on peaks ticked. now compress it multiple times until it is at a reasonable level, good way to check is the drop down arrow on tack and change view to waveform (dB) but default view is fine. Listen and look at your graph after every compress.

Sometimes its ready to roll then but sometimes ill norm to see if it sounds if i don't like it undo re comp then norm again.

Export, back into SVP, replace the audio track there with the new one then list with both game audio and voice over if struggle to hear vo the drop game audio a few dB (should take a lot).


DAMN I WISH THIS SITE GAVE KARMA CAUSE I THINK I DESERVE +1 FOR ALL THIS?!
Hey I read through everything you posted and first off I want to thank you immensely for taking the time out to tell me all of this. It helped me a great deal and I don't think any of what you said was harsh.

I went ahead and took your advice, fiddled with the settings and found that my microphone is pretty sensitive (overly so) when it comes to recording. it can pick up sounds coming from the first level of my house when its set to 100, so I sat about a foot away from the mic and set the volume to 10/100 in audacity and talked in a normal speaking voice and averaged about -16db, never passing -6db when 'yelling'. I will tweak volume based on the sound conditions around my house and environment.
 
D

DJStitch91

Guest
you asked this..
Audio wave forms are an enigma to me. How you can display and measure sound is a bit of a abstract concept to me.

My follow-up question(s) are pretty much related to wave forms and at what level you should average them at.

As you said earlier in your post you compress and then clip. According to the Audacity website, increases the sound of your audio (amplifies it) at a certain point and compresses the rest lower.

1. So does compression compress the loud sounds and then raise the quiet ones?

2. What does the threshold mean? Does -12 mean thats when it starts to boost things? -12 and lower? At a ratio of 2:1? So it doubles the lower sounds and quiets the loudest sounds by half?

3. Compression would make the background noise worse then, correct? So that is why I would want to use noise reductions first, correct?

4. Then after you remove the background noise you normalize, which according to the wiki normalize sets the peak amplitude to a default of -1. Why would you normalize at all? I don't understand. Why do you normalize?

5. What exactly does normalize do? It says sets the peak of the audio wave form to -1db (to prevent distortion and clipping), but then that means that it cannot be amplified further unless compressed. Does it force the other audio to scale uniformly. Example: My average audio is -12db, I yell and it causes my db to jump to -4db. I normalize to -1db. Does that mean it takes the loudest part of my audio, which in this case is -4db and bring it UP to -1db? Then does that take my average db of -12 and raise it to -9db (since -4 to -1 is an increase of 3db)?

6. Then you stated you compress it. Do you do this to try to even out the waveform and try to average the db of the entire clip? How many times should you compress it? You stated that you compress it until it looks and sounds right, but what does that mean?

7. To better understand this all, what average db do you think the entire audio clip should average in for youtube broadcasting / upload? -12? -1?

8. The next step, you said was to normalize again, sometimes. Is this always the last step, if so why? Do you want to bring up the entire waveform to an average of -1db? or close to it?

The main issue I'm having now is what average -db my entire waveform should be after compression and normalization to be joined back in with the video for broadcasting. Once I figure that out I will be able to join my game audio and either decrease the average db below my voice OR use auto ducking to force it to lower when I speak and then return to the same db as my voice when I'm silent.
thought it would be good to have answers here as well:

Some of these questions are pretty in depth of what you want to know, I'm not a sound engineer this is just stuff i have picked up from countless hours of fiddling and watching youtube videos.

1. So does compression compress the loud sounds and then raise the quiet ones?
Remember the tick boxes at the bottom? this is what they do they choose how they effect the audio. the one left box tick generally leaves big peaks big and brings the little peaks higher, that's my understanding of it anyway. By selecting both i believe (going by memory) big peaks shrink, leak peaks rise.

2. What does the threshold mean? Does -12 mean thats when it starts to boost things? -12 and lower? At a ratio of 2:1? So it doubles the lower sounds and quiets the loudest sounds by half?
Threshold: The level above which compression is applied to the audio <-- audacity site
I think, i may possibly be wrong, it means anything below this level gets worked. The last bit of this question is answered above again.

3. Compression would make the background noise worse then, correct? So that is why I would want to use noise reductions first, correct?
Yes and kind of, I like cleaning audio first, sometimes you will compress and find more but like i said last time game audio will cover most of these little bits. If you listened to just my voice over then sometime you can still here little bits, you need to balance abusing the audio and getting background down enough as there is a point where noise reduction will affect the rest of the audio.

4. Then after you remove the background noise you normalize, which according to the wiki normalize sets the peak amplitude to a default of -1. Why would you normalize at all? I don't understand. Why do you normalize?
Not everything is a matter of graphs and numbers, I do it by ear of what sound best through 1 my headphones, 2 my tv and 3 if i have time my gf's tv/headphones. But i don't shout a lot in my videos so I like all my audio levels level, above the game audio.

5. What exactly does normalize do? It says sets the peak of the audio wave form to -1db (to prevent distortion and clipping), but then that means that it cannot be amplified further unless compressed. Does it force the other audio to scale uniformly. Example: My average audio is -12db, I yell and it causes my db to jump to -4db. I normalize to -1db. Does that mean it takes the loudest part of my audio, which in this case is -4db and bring it UP to -1db? Then does that take my average db of -12 and raise it to -9db (since -4 to -1 is an increase of 3db)?
I just had to test this myself and yes? I pretty sure that is what it does.

6. Then you stated you compress it. Do you do this to try to even out the waveform and try to average the db of the entire clip? How many times should you compress it? You stated that you compress it until it looks and sounds right, but what does that mean?
I compress it to try and amplify it without amplifying the already loud enough bits.

View attachment 1127
This is a screenshot of 4 tracks i have saved track 1 is VO1 (voice over ep 1) t2: VO1 Edited
t3: vo2 t4: vo2 edited.

t1/3 are complete raw inputs, t2/4 are the final product, all the peaks of speech are within so many of each other so its sounds all near the same volume level, so when i play it back it all sounds the same level no loud bits or quiet bits as the that's what i want from my audio, (also helps with so many of us with different mic levels)

7. To better understand this all, what average db do you think the entire audio clip should average in for youtube broadcasting / upload? -12? -1?
well I normalize to -1 as you can see above. (who ever said i have understanding was lying haha I've just picked up bits and pieces that I think and hope are right.) But to put this in to context my commentary is about -1 the game i usually drop lower by 5 dB (depends how many explosions there are) sometimes i have to drop louder game audio in vegas lower by cutting the loud bit either end extending a few seconds to fade it in and out to that level and then drop just that piece of track until I can clearly hear Voice over the loud noise, although thinking about it now a better way to do that would be drop the game audio into audacity and normalize it with a -5 to -10 limit? which I'm going to try next time. (see it just about time, trial and error)

8. The next step, you said was to normalize again, sometimes. Is this always the last step, if so why? Do you want to bring up the entire waveform to an average of -1db? or close to it?
When you compress sometimes its can go higher and clip or you will listen to it and it just doesn't sound good so normalizing brings it back to the -1 level and hopefully it sounds better.

The main issue I'm having now is what average -db my entire waveform should be after compression and normalization
Let's say -1. tweak if needed.

Once I figure that out I will be able to join my game audio and either decrease the average db below my voice OR use auto ducking to force it to lower when I speak and then return to the same db as my voice when I'm silent.
Personally option 1, option 2 would annoy me as a viewer.
 
Last edited by a moderator:
Top