Fledgling voice artist seeking counsel

Greetings, all.

I’m half-way through a voice course but I’ve had to pause it temporarily so please don’t pay much attention the voice itself but I’m hoping there might be someone kind enough to assist me with my settings.

Edit - (Sorry, moderator, I left out some rather important details I see now).

Audacity 2.0.5.
Windows 8.1 (with update)
Snowball Blue microphone with shock ring and filter.
Home-made mini-booth which surrounds the microphone (I still have to wait until late at night for less traffic noise, etc. though until I can turn my closet into a studio).


I’ve been browsing for a while and I’ve picked up quite a few tips from here plus the requirements for the Audible.com site.

I think I have the list of steps I need now. I added to the start of it tonight after reading the wonderful advice here: https://forum.audacityteam.org/t/does-noise-removing-twice-badly-impact-the-actual-rec/30078/1

This added to my list amplify, settings for noise removal and the wonderful Noise Gate tool.

I then proceeded with what I already had but ended up doing another noise removal at the end.

I quite like the result. It involved a lot less work in removing extra noise. I think the background hiss is perhaps a TAD loud but I agree that dead silence is likely not appreciated by most listeners. One think I would like to fix, however is a strange wavering/wobbling every now in the dialogue. I promise I’m not wavering and wobbling myself. :slight_smile:

If there is anybody experienced who would take a look at my list and a listen and let me know what I should or shouldn’t be doing I would very much appreciate it.


The list:

  • Audacity tips
  • Always record silence at the start for the noise removal.
  • Amplify on default (0 db).
  • Remove noise.
    10
    6
    1000
    0
  • Select conspicious unwanted noise between phrases and silence them (CTRL+L).
  • Noise gate.
    Gate.
    Link stereo.
    No.
    0.0
    -22
    -50.0
    250.0
  • Normalize at -2dB.
  • Compress at 18dB, -
    50dB noise floor.
  • Equalization base
  • boost and treble
  • boost.
  • Normalize again at -
    2dB.
  • Hard limit at -3dB.
  • Normalize for final time at -2dB.
  • De-ess if needed for any sibilant sounds.
    7k
    Stereo
  • Convert to female:
    F#/Gb up A#Bb
    Semitones: 4.00
    Percent change 13.290

And the file as it stands now:

https://onedrive.live.com/redir?resid=FF7251924FF7642D!42773&authkey=!AEacDqviT__VLho&ithint=folder%2c.wav

Cheers.

Too much work. That’s what happens when you do too many facelifts and botox shots on the sound. Not only wobbling, but the voice sounds like “talking down a tube.”

It’s a danger sign when someone posts about the second time they applied noise removal. Generally, the first one causes enough troubles.

What was wrong with the original monolog? Post some of that, say the first minute or so. No filters, no effects, and no corrections. You did save the original live monolog as a WAV file, right, or maybe a Project, although that can be dangerous.

~~

Conversion of your voice to a different kind of voice (male/female, adult/child, etc) has never been something to call the papers about. Some of the best “conversions” involve acting, not software. Bart Simpson is a woman.

Not all mouth sounds change when you change people. Sibilants and mouth clicks tend to stay the same between people even when the pitch of the voice moves. Effect > Change Pitch changes everything, so it has the effect of dragging your finger on a record. Not natural at all. Two or three semi-tones may be the best you can do without the voice starting to sound funny.

We usually recommend making nice with a woman and let her do the female parts.

Koz

Thank you for the reply, Koz. :slight_smile:

That was at the bottom of my list and was just an “in case” sort of thing that I had picked up on the forums. I’ve not actually used it but then I’ve not reached the parts of the book where I have a lot of female dialogue. You don’t want to hear me as I follow the training videos on sounding like a female. hehe.

The good thing is that after I played with character voices and actual American accents for the dialogue, the author told me that it might be a tad too jarring for the listener because my voice was so different to my narration voice (which is what he really likes), so I’m now able to tone it down for all the differences.

As for the settings, well I was afraid of that. I already had a lengthy list just picked up from here and the ACX submission requirements: http://www.acx.com/help/acx-audio-submission-requirements/200485520.

I’ve not done a second noise removal before but this time (presumably because of far too much face-lift work) I found it necessary.

Also, I’m such a fool (although it was 04:00 in the morning) and didn’t save a raw version. I shall do one for you, though and put it in the same place later tonight.

Thank you.

You don’t want to hear me as I follow the training videos on sounding like a female. hehe.

I did a passable woman in a theatrical piece once. The producers were resigned to having to overdub a woman in post, but I leaned into the microphone and did a good enough Lauren Bacall that they left it in. Lauren Bacall’s voice was lower than Humphrey Bogart.

“You know how to whistle, don’t you Steve?”

just picked up from here and the ACX submission requirements:

There is a spirited discussion about the ACX requirements and I have one of the ACX senior elves on email. I want them to give me a narration of how they do quality acceptance. You “spoke” to an actual human, right? We want to know how much of this is algorithmic testing and how much is it is a comfortable old aunt sitting in a corner with good headphones. She wouldn’t have to be in an office, would she? She could be a Miss Marple in the garden being served strong tea by Cherry.

…but this time …

We’re finding USB microphones kill people trying to do this. "You need to get your noisy computer further away from your microphone — as long as you don’t exceed the 3 foot (1M) USB cable restriction.
!@#$%^&

didn’t save a raw version.

AAAAAAAAAAA! [breaking dishes sound]

What are you going to do when the computer goes for a noise removal process and just doesn’t ever come back? Three hours of work down the loo. Recovering a damaged Project is not for the easily frightened.

~~

We are messing with the idea of a standard suite of tools so people don’t have to do what you did, guess at it from the ACX guidelines. What did you think when you got to the specs that called out dBs and RMS? Did your Eyes Glaze Over? Oddly RMS (roughly loudness), although perfectly defined, isn’t easy. Audacity doesn’t have native tools to measure that. So even if you do everything right, you’re flying blind on some of these tests.

And we get some of the presenters whose submissions fail. Although many times we’re not as in the dark as they are, we’re all stumbling around in serious twilight.

That makes us angry. You wouldn’t like us when we’re angry.

Koz

Koz,

what a way with words you have. By your use of the word “loo”, however, I discern you must either be British (as I am) or have been heavily influenced. Either way should grant you an advantage when it comes to verbosity. :wink:

Lauren Bacall, eh? Well he did say he wanted the first female voice (who comes across on a radio) to sound like the lady in the original “Fog” film, that husky, sexy sound. I can’t do that for all of them, though. heh. Plus I have to do some sentient zombie voices (though I’m rather looking forward to those). Oh and yes, you really ought to read Joseph Duncan’s work. Start with either “The Oldest Living Vampire” as I did, for the vampire oriented, or “Mort” for the zombie oriented. Just don’t read Cattle if you’d like to hear my finished audio version (assuming I ever get there).

Right, so, apologies, I arrived home later than expected so I was only just able to throw together a raw version. No voice warm-up, no green apple, nothing! So don’t listen to him rabbiting on, I just wanted you to hear the ambient noise. I do usually wait until even later as I’m not too far from a main road here in Los Angeles so you can imagine how much traffic there is. And grrr, that police helicopter which circled for almost two hours a few nights ago did not leave me best pleased.

Hmm, I should post pictures of my setup. I have a home office so I put up some sound-proofing on the wall behind me and bought a big plastic carton thing with a lid and latches. I glued foam mattress topping to the inside and cut a hole in one side for my microphone stand to come through. It actually is only about two feet from the laptop but this is a very quiet machine (my other desktop beast has to be shut down). Anyway, it could give some good inspiration for others like me starting out on a low budget. Eventually I’m going to move it all into a rather large closet I have in the hall (which just doesn’t have any power sockets or ventilation gulp).

As for ACX, I only went with what was on the web-site and haven’t spoken to any Elves. And yes, some of them I had to learn as best as I could from here and elsewhere.

I’m surprised and disheartened to hear that some of your presenters get rejected. That isn’t instilling much confidence over here, matey. But no, I’d rather not see you when you’re angry. :wink:

So without further ado (and rambling), the new, raw version is in the same folder awaiting your keen senses.

Cheers.

I’m surprised and disheartened to hear that some of your presenters get rejected.

You should probably be reheartened. The people who make it here many times have trash and are First Time Outs depending on Noise Removal to Solve All Their Problems. Nope. Sorry. Can’t take out the dog barking. Wish we could. Nice dog, though.

You can’t violate any of these rules:

The Four Horsemen of Audio Recording (reliable, time-tested ways to kill your show)
– 1. Echoes and room reverberation (Don’t record the show in your mum’s kitchen.)
– 2. Overload and Clipping (Sound that’s recorded too loud is permanently trashed.)
– 3. Compression Damage (Never do production in MP3.)
– 4. Background Sound (Don’t leave the TV on in the next room.)

Then, after you make it that far, you have to conform to ACX requirements. Oh so many dB this way and oh so many RMS that.

Then, we assume you have to sound human and interesting. My objection to the posted reading is the voice starting to sound metallic and honky.

just able to throw together a raw version.

There’s no great rush. Nobody is holding a watch.

I discern you must either be British (as I am) or have been heavily influenced.

I’m bilingual. I know to ensure my mains are earthed under the bonnet of Lori (my lorry) in the car park before making off to the Tesco.

not too far from a main road here in Los Angeles



I live where Pico and Olympic meet.

Koz

boggle

I just out loud said YOU ARE KIDDING ME!

I live on South Orange, near where La Brea & Olympic meet! We’re practically neighbours! You should come by. :slight_smile:

Wow. Small world, eh? hehe

Excellent four horses tips there. I hope I’ve not shown you that I’m breaking any so far and I think I have a somewhat decent voice (according to Edge Studio’s initial assessment and the author I’m working with anyway). :wink:

I shall take a look at your pictures now. I just had to reply right away as my gast was well and truly flabbered when I read your response. :slight_smile:

Here are some pictures of mine at my new business page: https://www.facebook.com/media/set/?set=a.1391519691122274.1073741830.1390748814532695&type=3

Forgive the shoddiness of the logo and header. It is merely a placeholder and I’m working on a far nice and more original one. I got two clients (a social media management client and the author) before I even knew I was starting a business. heh.

Were you able to download the raw version?

If you’d care for faster communication my e-mail is XXX@XXX.com and FB is http://www.facebook.com/ian.m.walker

Cheers.

Oh, that’s revealing, isn’t it? I need to sit and listen to it critically. Did we establish what kind of microphone you had? I think the “honky” sound is your studio. I couldn’t tell before with all the rouge, powder and lipstick on top. Oh, and you have an extraordinary volume problem.

As we go.

Koz

What do you think to this version?
https://www.dropbox.com/s/kjurco8kpvmfedz/Chapter%2001%20(raw)%20-%20after.mp3

Yes, Blue Snowball.

Haha, there is no rouge, powder or lipstick in this house, mister. Volume problem? I record at only half volume and then followed the advice to amplify, etc. :frowning:

And it was you I just accepted a friend request from I hope. I usually vet someone I don’t know but assumed it must be you.

Cheers.

Drat, I ought to have used multi-quote. This whole “every response needs to be checked by a moderator” is already becoming rather tiresome.

I shall have a listen right now, Steve. Thank you.

Cheers.

Steve,

wow!

No background noise whatsoever and I sounded natural (at least to my untrained ear).

It was also excellent to hear myself before my rigmarole of warm-up, chugging tons of water and nibbling on my green apple as I go. I can now see it is well worth it as it makes a huge difference. In this version I could hear all the clicks of my mouth. Horrible. I can’t resist asking, however, what you thought of the voice in general for someone narrating a book. Is it one you would listen to?

So how did you achieve this magic please? :slight_smile:

Cheers.

Oh and can’t you please change the setting that is forcing moderator view of every reply? Usually a forum no longer requires such once one has joined and the moderators are only pestered when a post is reported. Surely that would be far friendlier and less work for the staff?

Hi MichloIW

That’s much better than the first version, which has much too much bass.
In this version too, there’s some boominess to the low frequencies, probably due to the recording environment.

The recording level is a bit low (-23 dB, actually -26 dB, since the highest peak comes from taking off the headphones), you give away nearly one half of the bit depth (16 bit minus 7 bit), aim for -6 dB. Also, record only one channel.
You can measure the RMS value with “Contrast” in the analyse menu.
It is currently fairly low (-27.1 dBFS) and only ittle about the noise floor (-40 dBFS). This makes it harder for the noise reduction to work properly.

Nice version, Steve.
A pity that there’s sometimes underlaying traffic sound (e.g. at 1:38).

As Robert wrote, the recording level could do with being higher.
Here is a section of your recording (upper track) and what I would think of as an ideal recording level (lower track). For clarity I’ve converted the track to mono.
tracks003.png
Note that, other than the “clunk” when you take your headphones off (which is not part of the “real” recording), the biggest peaks are around half of the track height (-6 dB).


I would recommend that you record in mono (set the number of channels to “1 (Mono) channel” in the device toolbar).
Your microphone is mono, so there is no advantage to recording 2 channels.
Recording 2 channels has the disadvantages that it creates twice as much date (bigger projects, more work for the computer and so more chance of producing clicks).
If the final format is to be MP3 (or other compressed file format), mono will give better sound quality for the same file size.

Perhaps you need better headphones :wink:
My main complaint about this recording is that even after applying as much noise reduction as I dare (without making the sound go metallic and bubbly), the noise level is still a bit high.

The noise problem is mostly environmental (traffic). There’s even a (police car?) siren going off at 3:55.

I was wondering if the “sound proof box” that you have the microphone in is contributing to that.
It would be interesting to hear a side by side comparison of a short recording with / without that box.
If the box is adding to the boominess, then adding some bracing to make the sides more rigid may improve things.


by doing as little as possible :wink:

Processing audio can never put back anything that is missing, it can only remove or exaggerate what is already there. You are never “adding quality”. The best you can hope for is to reduce the distractions. Starting with as good a recording as possible is the vitally, overriding, most important thing. (Have I emphasized that enough? :slight_smile:)

These were the steps that I took:

  1. Convert to mono (no point wasting disk space or processing time on a mono recording).
  2. Delete the “headphone clunk”. We don’t want that.
  3. Normalize to 0 dB (with DC offset correction) Audacity Manual I like to be able to hear clearly what I am doing.
  4. High Pass filter. 80 Hz, 12 dB per octave. This is essentially a “rumble filter”. Note that filters do not cut off dead at the specified frequency, they “roll off” beyond that frequency. Audacity Manual There is not much “voice” below 80 Hz, it is mostly noise, so we don’t want it. I prefer rolling off fairly gently (6 to 12 dB per octave) to avoid an “unnatural” sounding cutoff.
  5. Noise reduction (using the “Noise Removal” effect. This was the trickiest part - getting as much noise reduction as I could without making the voice sound unnatural. Watch out particularly for “s” sounds becoming “swishy” and consonant “reverberating”.
  6. Applied a “limiter”. NOT a “hard clipping” limiter. A proper non-distorting limiter. Unfortunately Audacity does not currently ship one of these so you will need to use a plug-in (add your vote here if you want Audacity to include a limiter as standard: Adding Features - Audacity Forum
    This reduces the level of the highest peaks, allowing the audio to be reasonably (but not overly) loud without excessively high peaks.
  7. Final amplification to desired peak level (-3 dB is the ACX requirement).
  8. Export as WAV. I didn’t actually do this, but for production work this is an essential back-up step.
  9. Export as MP3. 192 kbps CBR (as per ACX requirements).

Notice that there are few actual processing steps. Mostly just noise reduction and “compression” - in this case “limiting” because the voice level was nice and even, so I only wanted to reduce the highest peaks. From a technical perspective, other than the background noise (and the low level), there’s not much wrong with the original (congratulations).

It could perhaps do with a small adjustment to the Eq to reduce boominess, but it would be better if that can be improved in the source recording.

Yes, low volume and there’s something about your environment that’s causing slap. That’s the talking down a tube thing I complained about at the beginning. That’s causing the bass boost similar to sticking your head in a barrel boosts one tone where the barrel “rings.”

I call it slap because it’s half-way between pitch change effects and full-on echo. I suspect if you hit two pencils together I could tell you how far away the wall or desk was.

Do you have your Snowball set to directional (cardioid)? You should not be using the 10dB attenuator. How big is the room? Do you have it on its little stand? I probably wouldn’t do that. Note how high up the microphone is in the next picture.

In this setup the room is soundproofed, but I put a double moving pad on the desk to get rid of slap effect.
http://www.kozco.com/tech/audacity/pix/JMASoundShoot.jpg

If you have anything close to the Snowball, you may be interfering with its internals. Directional microphones need access to both the front and back.

I know there is the temptation to try and get rid of the odd effect with equalization tools, but room problems are not so simple. You run the risk of producing an “equalized barrel” instead of clear speech.

To put those numbers in perspective, sound level doubles and halves every 6. So when I had to boost your raw presentation 30dB, that’s 6, 12, 18, 24… That’s just too low. Peaks in -6 to -10 are good for a raw, live performance.
http://kozco.com/tech/audacity/Audacity1_record.jpg

If you have an openly hostile environment (La Brea and Olympic), then I would investigate a head-mounted microphone, or borrow a trick from This American Life and use a shotgun microphone close-up. I don’t think a Snowball stands a snowball’s chance of producing a quiet, well behaved recording in that neighborhood. Try a head-mounted directional microphone (or head_set_). Most gamer headsets are pretty awful, but Logitech makes headsets that do OK.

I’ve used a Snowball. It’s not a dreadful microphone, but it’s not a gift from the angels either. It does like a quiet room and free access on all sides.

In case nobody has said yet, I love the presentation. I could listen to that on long walks — if it didn’t sound funny. No, I don’t think you could fake a woman’s voice. My presentation was an accident.

I need to drop out for a while.

Koz

You passed the “Brit Test.”

Koz

There’s a box-gadget with lined pointy acoustic-foam which reduces the reflections from the room …

It’s a bit expensive for what it is but cheaper than soundproofing the room. [ the box could be improvised ].

Thank you, everybody.

So much to take in.

First things first. Robert, the background noise you hear (from the hostile environment of La Brea as Koz describes) I left in because this was RAW plus I was wondering if any skills could show me how to deal with it if I do miss it and record it. Usually I have to listen out for it, pause and / or re-do when I hear it. This is also why I have to start late at night until I can afford to convert my large hall closet. This is also why I took the headphones off as I really wanted to listen out for it but leave it in this time.

So, the recording level, Steve. Perhaps I’m too far away from the microphone as well because I thought I needed to have the microphone completely enclosed by the mini booth I built. It is on a microphone stand which I put some padding over, BTW.

I’ve added three new pictures to my home studio album here:

https://www.facebook.com/ian.m.walker/media_set?set=a.10152353204506411.1073741847.641111410&type=1

The first shows a panoramic view. The large monitor would be off when I’m recording because that is an old desktop machine which makes too much noise and has to be shut down so I work from the almost silent laptop. The window behind would also be closed, of course.

The second shows the padding I added around the microphone arm once I decided I needed to put a hole in the box so that the microphone could be better enclosed.

The third shows about how far I usually am when I’m standing to record. Is that too far?

I shall go back to recording in mono. I saw on another post someone say “people these days usually prefer stereo” which is why I made that change a few days ago.

I shall try to do a recording with the microphone as far outside of the box as I can in order for your comparison, Steve.

Haha and yes, you have emphasized enough that the recording has to be good in the first place. :wink:

I’ve noted down your new steps. Could you confirm which settings you use for the noise reduction please. And sorry about the hard limiter, another thing I thought I needed for the dastardly ACX. I shall definitely vote a yes on having a Limiter included (edit - though I’m having trouble finding the actual post). In the meantime, is there one you would recommend please?

So I really don’t have to do any compression or equalization? Nice. That means my voice gets to sound more like “me”.

Aww thank you very much for the congratulations. That actually means a lot to me.

Koz, yes the microphone is set to Cardiod. At a rough guesstimate the room is about 21 x 11 ft. I know, too big but I can’t yet move into the closet (ooer) until I can figure out power and what not.

I too have padding under the whole box and I have some underneath my laptop. I even have it around the music stand which holds my tablet. Plus I’ve seen that picture before so it seems I’ve been reading about you longer than I realized. :wink:

As to what is close to the Snowball, I did worry about the music stand. I can’t put it underneath because a) it wouldn’t fit and b) we were taught it is best to be looking up a little whilst reading. I have it as close as I can to where I don’t have to completely face the tablet as I worried that would cause some bouncing of my voice from the glass of the tablet. Perhaps I can figure out a way of having atop the box instead so that the space in front of the microphone is clear and I really am looking up.

Hmm, I can’t afford a new microphone right now (new business after 2 years unemployed and burning through most of my savings before finally getting my arse in gear means having to be careful with the pennies). I do have a Logitech G35 headset which was a gift from my sister but I’ve always disabled that microphone. Are you saying it might be better than my precious little Snowball? :frowning:

And much gratitude to you too about the presentation! Seriously. And hah, so I can’t fake a woman’s voice, eh? With the videos I’ve been watching (mostly trans women teaching others how to find their female voice) I might agree though I’m still keen on trying just for the hell of it. Each of those say one has to start at one’s highest falsetto (I swear I hear cats screaming and dogs howling when I do it) and then drop down just a little to a more “comfortable” range. Then add breathiness and some sing-song quality. Sounds like it is a lot of work for females to speak. Odd they seem to do it so easily. :wink:

And I passed a Brit test? I didn’t even know there was one! Is there a reward? :slight_smile: Edit - Ooh, could it be that my message was posted directly without needing the great approval from up above? :wink:

Trebor, I quite like the padding inside of that box but it all seems very close to the microphone and a tad rickety. After seeing the foam inside my home-made version would you recommend I change the type of lining I have?

Thank you very much, everybody. You’ve really lifted my spirits and given me a much more hopeful feeling that I can get this nailed down. My author really wants us to get the first book done and out there. I have the great fortune of him wanting me to do his others after that. :slight_smile:

Cheers.

You can make mono into pseudo-stereo , e.g. with Steve’s Audacity plug-in … Missing features - Audacity Support