Hi Forum
https://librivox.org/uploads/tests/VOgalUS_sample.mp3
I have been doing recordings for a while and would like to know how I sound. This is a sample. Its a nonfiction narration and is a raw file. I'd like to know what you may hear in the background and anything about my voice - siblance/mouth noise... Any suggestions or comments are welcome.
Audio sample for review
Hey, it sounds good! No real mouth noises and equipment noise is there but not too bad and can be addressed. I am guessing mic is at a good distance and room sounds pretty dry with no real problem reflections.
I went ahead and did my usual processing, eq, compression, de-ess and expansion then uploaded the result to my dropbox so as not to burden LV resources:
https://www.dropbox.com/scl/fi/5nl4j3um8ei0fuccou4e4/vogal-test.mp3?rlkey=w4bbmlzvmxi5bicgfw9eeysxy&dl=0
The only curious thing as to the sound was in eq-ing I noticed some variation to the frequency response in the original. Was it processed at all? It could be also that mic position may be to the side where head turns away and back toward the mic can cause this. It could also be in the hardware signal chain (mic and/or interface). However, it is very slight and only noticeable listening with good headphones/amp.
I did the processing with the Sheps Omnichannel plugin by Waves because it is easy to see all at once. Here is a screenshot:
https://www.dropbox.com/scl/fi/xaqjocx3909b5ynclwbj5/Screen-Shot-2023-08-07-at-7.42.33-PM.png?rlkey=tb9jm8b9ruo7klbgnxk5d2z2a&dl=0
In the Sheps I also gained up a little (~4.5 dB overall over the original) and used the expander to keep noise low. Basically, everything done was a low cut filter, a parabolic eq notch in the high-mid region, VCA compression (a little stronger than I usually use but not too heavy), de-ssing mostly in 5 kHz region, a bit of expansion to tamp down the noise area that got higher when gaining up the overall levels.
Of course, for Librivox, this isn't really necessary but your sound has a lot of potential and if you are interested in the engneering/mastering side....
The only other thing that would be helpful to know for someone trying to help is what equipment you are using, details of recording space including mic placement, and software used and if/how processed.
I hope this helps somewhat; I only did it as a "proof of concept" in helping Librivoxers who might be interested in improving sound quality.
Regards,
scotty
I went ahead and did my usual processing, eq, compression, de-ess and expansion then uploaded the result to my dropbox so as not to burden LV resources:
https://www.dropbox.com/scl/fi/5nl4j3um8ei0fuccou4e4/vogal-test.mp3?rlkey=w4bbmlzvmxi5bicgfw9eeysxy&dl=0
The only curious thing as to the sound was in eq-ing I noticed some variation to the frequency response in the original. Was it processed at all? It could be also that mic position may be to the side where head turns away and back toward the mic can cause this. It could also be in the hardware signal chain (mic and/or interface). However, it is very slight and only noticeable listening with good headphones/amp.
I did the processing with the Sheps Omnichannel plugin by Waves because it is easy to see all at once. Here is a screenshot:
https://www.dropbox.com/scl/fi/xaqjocx3909b5ynclwbj5/Screen-Shot-2023-08-07-at-7.42.33-PM.png?rlkey=tb9jm8b9ruo7klbgnxk5d2z2a&dl=0
In the Sheps I also gained up a little (~4.5 dB overall over the original) and used the expander to keep noise low. Basically, everything done was a low cut filter, a parabolic eq notch in the high-mid region, VCA compression (a little stronger than I usually use but not too heavy), de-ssing mostly in 5 kHz region, a bit of expansion to tamp down the noise area that got higher when gaining up the overall levels.
Of course, for Librivox, this isn't really necessary but your sound has a lot of potential and if you are interested in the engneering/mastering side....
The only other thing that would be helpful to know for someone trying to help is what equipment you are using, details of recording space including mic placement, and software used and if/how processed.
I hope this helps somewhat; I only did it as a "proof of concept" in helping Librivoxers who might be interested in improving sound quality.
Regards,
scotty
Hi Scotty,iBeScotty wrote: ↑August 7th, 2023, 8:57 pm Hey, it sounds good! No real mouth noises and equipment noise is there but not too bad and can be addressed. I am guessing mic is at a good distance and room sounds pretty dry with no real problem reflections.
I went ahead and did my usual processing, eq, compression, de-ess and expansion then uploaded the result to my dropbox so as not to burden LV resources:
https://www.dropbox.com/scl/fi/5nl4j3um8ei0fuccou4e4/vogal-test.mp3?rlkey=w4bbmlzvmxi5bicgfw9eeysxy&dl=0
The only curious thing as to the sound was in eq-ing I noticed some variation to the frequency response in the original. Was it processed at all? It could be also that mic position may be to the side where head turns away and back toward the mic can cause this. It could also be in the hardware signal chain (mic and/or interface). However, it is very slight and only noticeable listening with good headphones/amp.
I did the processing with the Sheps Omnichannel plugin by Waves because it is easy to see all at once. Here is a screenshot:
https://www.dropbox.com/scl/fi/xaqjocx3909b5ynclwbj5/Screen-Shot-2023-08-07-at-7.42.33-PM.png?rlkey=tb9jm8b9ruo7klbgnxk5d2z2a&dl=0
In the Sheps I also gained up a little (~4.5 dB overall over the original) and used the expander to keep noise low. Basically, everything done was a low cut filter, a parabolic eq notch in the high-mid region, VCA compression (a little stronger than I usually use but not too heavy), de-ssing mostly in 5 kHz region, a bit of expansion to tamp down the noise area that got higher when gaining up the overall levels.
Of course, for Librivox, this isn't really necessary but your sound has a lot of potential and if you are interested in the engneering/mastering side....
The only other thing that would be helpful to know for someone trying to help is what equipment you are using, details of recording space including mic placement, and software used and if/how processed.
I hope this helps somewhat; I only did it as a "proof of concept" in helping Librivoxers who might be interested in improving sound quality.
Regards,
scotty
Wow. I can't believe the improvement. You are using a plugin that I can't get yet and is far more advanced than my knowledge of mastering. I am definitely interested in improving sound quality. Heres what I use:
Audacity
Rode Nt1 5th generation mic
Scarlett Solo interface
PC laptop
Sample wasn't processed
Fort-like studio using moving blankets
Can you recommend effects I can use in Audacity? I have gotten familiar with them since I started reading for Librivox and using Audacity about a year ago. The free realtime effects are downloaded in my Audacity but I am not totally familiar with them. Or you might have some other suggestions.
Over the past year I have been working on my reading, increasing the length of the reads. I am doing around 1500 words right now in a sitting. I've also learned to edit out the breathing noise between sentences using room tone. The readings are getting better. When I get a little better I'll look at getting some coaching. Its the sound I'd like to improve right now. The equipment is new and I am working on my mic placement. It's nose bridge height and about 6" away a little bit off to the right hanging upside down. The pop filter has been removed.
Any suggestions are welcome. I heard about plugins at Waves and Izotope and have been recommended people who could create a 'plugin' to improve my sound quality. Maybe when I get more familiar with reading and mastering. This new mic is very different than the $10 USB mic I got when I started. I expect to have it for a long time.
VO gal
Do do do do, de da da da is all I want to say to you.
Do do do do, de da da da is all I want to say to you.
That is excellent equipment (I have always heard good stuff with the NT1 and used to have a 1st gen Scarlett 8i6 that was awesome) and the diy fort setup is really all you need, I think, because it does sound good and dry (for easy post processing). Honestly, I am not that familiar with Audacity but know it is very powerful and can use plug-ins. I don't think the particulars (of what software and plug-in is used) matters as much as the concepts. In the example I uploaded, could have gotten the same result with the built in plug-ins to Reaper.
My general philosophy for voice recording is
1.) get the cleanest possible sound going in with is little ambient and equipment self-noise reasonably achievable along with as little echoes, reverberations and resonances possible (I struggle with this in my own space). Getting the mic as close as possible before hearing too much mouth noise (and experimenting with mic angles) or detrimental proximity effect (also mic dependent) helps with this. Usually 6-8 inches is a good starting point for condenser mics, maybe a little closer for dynamic mics.
2.) Gain stage the recording, getting a raw signal that averages not too high, with plenty of headroom for post processing. Speaking normally the meters in my recording software would show around -12dB or less if I know I may get louder in parts.
3.) Process the signal first with some corrective equalization, usually cutting the low frequencies that aren't necessary for voice and sometimes aggressively if I have a lot of proximity effect that can be a problem with my own dull voice. Then finding and reducing resonant frequencies that my space is causing (a lot of trial and error). Then add some sweetening EQ maybe for more clarity with a high shelf of a few dB. After this I compress and gain up to make the voice sound more even. Now the straight linear gain will also increase the noise floor of the recording so I quiet this area back down with the expander (so called because it expands the dynamics of the signal but in the quiet region). Usually If I have added clarity, sibilance may also become more prominent so I address this with a de-esser which also requires some trial and error to find the right setting. Don't want to go overboard with this and sound like you have a lisp, haha. Lastly, I may gain up a little at the output stage if necessary with a limiter to prevent peaking and use a dc offset filter.
I would be very surprised if you couldn't do this processing in Audacity, however I don't know if Audicity allows non-destructive editing which is practically a necessity for mastering. I loved Reaper for this flexibility and back when I bought it, many moons ago, it was very reasonably priced at around $50 and they gave you an unlimited free trial.
All the best!
scotty
My general philosophy for voice recording is
1.) get the cleanest possible sound going in with is little ambient and equipment self-noise reasonably achievable along with as little echoes, reverberations and resonances possible (I struggle with this in my own space). Getting the mic as close as possible before hearing too much mouth noise (and experimenting with mic angles) or detrimental proximity effect (also mic dependent) helps with this. Usually 6-8 inches is a good starting point for condenser mics, maybe a little closer for dynamic mics.
2.) Gain stage the recording, getting a raw signal that averages not too high, with plenty of headroom for post processing. Speaking normally the meters in my recording software would show around -12dB or less if I know I may get louder in parts.
3.) Process the signal first with some corrective equalization, usually cutting the low frequencies that aren't necessary for voice and sometimes aggressively if I have a lot of proximity effect that can be a problem with my own dull voice. Then finding and reducing resonant frequencies that my space is causing (a lot of trial and error). Then add some sweetening EQ maybe for more clarity with a high shelf of a few dB. After this I compress and gain up to make the voice sound more even. Now the straight linear gain will also increase the noise floor of the recording so I quiet this area back down with the expander (so called because it expands the dynamics of the signal but in the quiet region). Usually If I have added clarity, sibilance may also become more prominent so I address this with a de-esser which also requires some trial and error to find the right setting. Don't want to go overboard with this and sound like you have a lisp, haha. Lastly, I may gain up a little at the output stage if necessary with a limiter to prevent peaking and use a dc offset filter.
I would be very surprised if you couldn't do this processing in Audacity, however I don't know if Audicity allows non-destructive editing which is practically a necessity for mastering. I loved Reaper for this flexibility and back when I bought it, many moons ago, it was very reasonably priced at around $50 and they gave you an unlimited free trial.
All the best!
scotty
I had to read this a few times to understand your process. Some of it applies to my recordings but I am not sure how it applies in Audacity. I am able to find some things online. Mic placement is what I am working on right now. Can you explain what a high pass filter is correcting? A professional audio engineer told me he would apply this to my recording. When I did the sound wave went very smaller. It also went quieter. Maybe I am not used to hearing what my voice should sound like. He promotes the least amount of editing as possible.
VO gal
Do do do do, de da da da is all I want to say to you.
Do do do do, de da da da is all I want to say to you.
Your recording sounds good.
A high-pass filter allows high-frequencies to pass through unaltered and removes low-frequencies. Your recording has some very low-frequency rumble. It's too low for human hearing, so it is good to remove it. I know it's there because it shows up when using 'Plot spectrum'.
Here is an EQ curve you can import into 'Filter Curve EQ' in Audacity. Copy the text below and paste into a .txt file. Then use the import option under 'presets and settings'
The EQ curve removes the low-frequency rumble and does some subtle adjustments to boost mid-frequencies and tame sibilance.
A high-pass filter allows high-frequencies to pass through unaltered and removes low-frequencies. Your recording has some very low-frequency rumble. It's too low for human hearing, so it is good to remove it. I know it's there because it shows up when using 'Plot spectrum'.
Here is an EQ curve you can import into 'Filter Curve EQ' in Audacity. Copy the text below and paste into a .txt file. Then use the import option under 'presets and settings'
Code: Select all
FilterCurve:f0="20" f1="72.380093" f10="12415.829" f2="395.71124" f3="708.4748" f4="1320.7972" f5="2713.3482" f6="4977.2674" f7="6092.8355" f8="8489.0318" f9="9979.7795" FilterLength="8191" InterpolateLin="0" InterpolationMethod="B-spline" v0="-57.461536" v1="-0.11140585" v10="0.058355331" v2="0.015914917" v3="1.3740053" v4="2.2228117" v5="1.7559681" v6="1.2042441" v7="0.35543776" v8="-2.5305042" v9="-2.4880638"Hi IainMF,IainMF wrote: ↑August 28th, 2023, 7:16 pm Your recording sounds good.
A high-pass filter allows high-frequencies to pass through unaltered and removes low-frequencies. Your recording has some very low-frequency rumble. It's too low for human hearing, so it is good to remove it. I know it's there because it shows up when using 'Plot spectrum'.
Here is an EQ curve you can import into 'Filter Curve EQ' in Audacity. Copy the text below and paste into a .txt file. Then use the import option under 'presets and settings'
The EQ curve removes the low-frequency rumble and does some subtle adjustments to boost mid-frequencies and tame sibilance.Code: Select all
FilterCurve:f0="20" f1="72.380093" f10="12415.829" f2="395.71124" f3="708.4748" f4="1320.7972" f5="2713.3482" f6="4977.2674" f7="6092.8355" f8="8489.0318" f9="9979.7795" FilterLength="8191" InterpolateLin="0" InterpolationMethod="B-spline" v0="-57.461536" v1="-0.11140585" v10="0.058355331" v2="0.015914917" v3="1.3740053" v4="2.2228117" v5="1.7559681" v6="1.2042441" v7="0.35543776" v8="-2.5305042" v9="-2.4880638"
Awesome. I'll give this a try. Learning audio speak is something I am also learning.
VO gal
Do do do do, de da da da is all I want to say to you.
Do do do do, de da da da is all I want to say to you.