Best voice sound settings

I recorded my voice using a specialized digital recorder, which saves as a wav file. It sounds great in Audacity and I edited with music under it for my intro. However, I’m doing a podcast on Blog Talk Radio. The intro music sounds fine, but my voice sounds like I’m Daffy Duck (“thuffering thuckatash”) or a strong lisp. I’ve had this issue recording before. However, my natural voice doesn’t sound like this and on the show it is ok.

I don’t understand what happens to the file between editing and uploading. I export as MP3.

Can someone direct me to where there are recommended settings for optimizing voice or give me some suggestions? I need to record an audio book in chapters and want the quality as good as possible.


If the Daffy Duck effect was on the original recording I’d suggest a “pop filter” in front of the microphone …

Ahh I see you are converting from a WAV to mp3.
If the bit rate is set low for this conversion the mp3 version can be noticeably poorer than the wav original.
You should use a bit rate of 256Kbs or even 320Kbs if you want the mp3 to be indistinguishable from the wav original.

In Audacity when you “save as file type as MP3” click on “options” (bottom right) then set quality to 256Kbs or 320Kbs, then click on “OK”, then “save” the MP3 file.

Of course the price to pay for higher quality (higher bit rate) is larger file sizes.

You might get away with a bit rate of 128Kbs if you sample/resample your audio at 22050Hz which is adequate for speech, (instead of the standard 44100Hz).

Trebor, that sounds like the answer, but it’s a bit over my head. :slight_smile: This is the stuff I need to learn! So thank you.

No Daffy Duck on the original recording. That’s what is so frustrating.

But even once I converted it, when I listened on my computer, the mp3 sounded fine. Again, it was only once I uploaded that this happened.

The wiki was very complicated. I wonder if there is an easier way to understand bitrates, sampling, and resampling?

Yes, I converted to MP3 because the wav file was huge.

I appreciate your help. I’ll try playing with those setting to test.

Thank you.

I’ve made an example of what too much compression (too low bit rate) sounds like.

If you click on this link you can download the before-after example mp3 …

If the type distortion in the second half of this mp3 is what you mean by “Daffy Duck” then your mp3 bit rate is set too low.

Below is the MP3 options menu in Audacity 1.3 where you set the bitrate when saving as MP3, (use 256Kbs or 320Kbs for high quality sound).
If you are using Audacity 1.2 you can alter the MP3 bit rate by clicking on “Edit”, “Preferences” (at bottom of drop down menu), then “File Formats”. The “MP3 export setup” should then be visible, again adjust the bit rate to 256 or 320Kbs for high quality*, click on “OK” after you have changed the bit rate. (* you could try saving at lower bit rates to see if the results are audibly acceptable, the lower the bit rate the smaller the file size and lower the quality of the sound ).

I’m using 1.2.6. Thank you for the clip. I’m not sure that is exactly it. Would you mind listening to the actual recording that I’m talking about? Here’s my link to the radio show:

If you click on the play button next to my picture, you’ll hear the Blog Talk Radio intro, then my intro music and then you’ll hear my voiceover (“Welcome to Best Friend Radio”). I think you’ll know exactly what I’m talking about if you hear it. :smiley:

Please believe me, I do not really sound like that.

Thanks for your help. I really want to get this right. I will study your settings as you described. I can upload the mp3 or wav vo file so you can hear that the file itself is vastly different than how it sounds on blog talk radio.

Thanks again.



from following your link, it soulds to me as though Trebor is right on the money - and you are indeed over-compressing your MP3s, so try a higher bitrate.

Also the signal soulds a little oversaturated - are you fetting clipping when you record - i.e. does the recording level meter go all the way ip to the 0 on the right hand side and leave the little red indicator bar there? BTW: you can get better control of the recording and layback levels by moving the meter bar and stretching it out - I have mine stretched across the whole screen width.

I thought I could hear you turning the script pages at times - you should either try to keep the script further away from the mic - or use Audacity to edit out or silence the page turns.


Hi. The only pre-recorded part is the intro over the music and that is my concern. The original recording is clear with no distortion and the recording meter was in the green zone.

The show itself is live on a cell phone, and yes, I’m still learning so there are pages turning. :slight_smile:

But my question is about the pre-recorded part only. Thanks for your help.

Ok, so for the intro you have two separate tracks in Audacity, one for an imported music track and one for you voice track to overlay the music - have I understood correctly?

  1. If so, when you play both tracks together in Audacity does the result sound ok?

  2. If you export the two tracks to a WAV file - does that resulting lossless WAV file sound ok?

It is possible to have individual tracks that are not over-saturated in themselves, but when you mix them in Audacity - on export say, then the resulting combined track can exhibit overload clipping.
You can test this in Audacity mixing the tracks nad then playing (in 1.2 use Quickmix and in 1.3 use Mix&Render)


I listened to your and other voices on and they are of similar sound quality: I suspect that Blogtalkradio is processing the audio you send them to reduce its file size and is consequently compromising quality when they do this, (so not your fault).

The noisy sizzle which follows your words sounds like an artifact of conversion to 8 bit resolution, (Audacity doesn’t , go any lower than 16 bit depth).
If your webcast provider is converting all broadcast material to 8 bit resolution this may explain your Daffy Duck lisp .

Again I have created a “before and after” this time downgrading a 16 bit resolution to 8 bit …

If Blogtalkradio are transmitting the audio you send them in 8 bit then you can avoid the Daffy Duck lisp/sizzle by reducing the high frequencies in the audio you send them to 5KHz (5000Hz).

I downloaded a “popular“ (Bill Phillips) webcast from Blogtalkradio and analysed its spectrum …
Popular Bill Phillips sound spectrum (5KHz cut off).png
The moral: don’t send Blogtalkradio audio frequencies above 5KHz, as they will be only be reproduced as sizzle noise.

Bit Depth
Bit depth defines how many bits are used to describe each of the samples taken from the sample rate . It describes the potential accuracy of a digital audio file. Higher bit depth audio will sound better than smaller bit-depth audio. 8 and 16-bit audio are currently the most common sample sizes. 8-bit audio takes up less hard drive space but is inherently noisier than 16 or 24 bit-depth audio. CD quality is 16-bit.

It’s not uncommon for these type of services to reduce the sound quality in order to reduce their band width.

Audacity can be persuaded to Export 8 bit audio if you really try, but internally it works at 32 bit, and most of the Export formats are 16 bit or more.

That looks like they are dropping the sample rate down to 11kHz (see “Nyquist”)

  1. If so, when you play both tracks together in Audacity does the result sound ok?

Yes, the result is fine. I did fade in and out with the music and it sounded fine.

  1. If you export the two tracks to a WAV file - does that resulting lossless WAV file sound ok?

The WAV file sounded fine too. When I export to mp3 file (which merged the tracks), that also sounded fine.

I guess I’ll have to study what everyone is saying to understand and just try these settings. I agree that BTR is reducing sound quality for bandwidth. It’s just that other people’s intros don’t sound like “Daffy Duck.” So they may be using these suggested settings.


Yes looks like they are dropping the sampling rate and the bit depth to reduce file size: dropping the bit depth produces the sizzle (see my example downrating to 8 bit above).

Agreed : Sbilheimer should resample her audio at a sample rate of 11025 before she sends it to BTR.

In Audacity 1.3 you can do this by clicking on “Tracks”, “Resample”, select “11025” from the drop down list and click on “OK” …
Audacity resample 11025.png

Thanks so much. I will work on this on Saturday and let you know.