Follow up from yesterday.

Firstly, big thanks to Kato for his suggestions.

The problem is, I've either tried, or actually use all of them... and while they help, they don't eliminate the problem.

From start to finish, here's what I do:

1) I record the podcast.

I try to get this in one take (and always fail) for pretty much one reason. As I've said before, my microphone is absolutely terrible. I can't stress this enough. If the microphone moves or changes the angle from my mouth by a single inch, you get a noticeable change in the audio.

As well as volume issues, I've noticed that even a slight change to that angle can make it capture a lot less bass, or make it sound tinny.

So I try to get it in as few takes as possible because if I'm reading continually I can minimize the amount that the microphone moves. Even if it does, you get a 'smooth' progression from one type of sound quality to another.

If I recorded each paragraph seperately (which is what I started doing) you'd hear me speak, then the next sentence would be a lot lower, followed by one much louder, followed by one that's tinny.

Like I said in yesterday's post, the sound dropped by over 10 decibels (I know, I checked) when I moved my head to the right by a half inch to look at the next page of the script. The mic really is that sensitive to movement.

Basically, I'd rather have the sound quality go up and down smoothly than change jarringly with each new paragraph.

The only real solution I see to this problem is sucking it up and buying a decent microphone.

2) Edit.

I go through my recording and edit as needed. Cutting out silences where I paused to clear my throat or cough, cutting the paragraphs where I flubbed my lines or stumbled over a word and generally making sure I have a coherent podcast where it at least sounds like I just talked from start to finish.

3) Mix down.

At this point, I've usually got anywhere from 4 to 10 different audio files in the timeline from where I cut the main file into pieces. (I'm using Adobe Audition, in case you're interested). So I mix down the vocal audio to a big sized, no compression .wav file.

4) Filters, filters, filters.

With the vocal for the podcast recorded, I run it through a bunch of different filters. I use a hiss reducer, pop and click eliminator, a slight bass boost and a normalize filter. The normalize filter helps, but the volume and sound quality can vary by so much, the distortion in the quiet parts can more noticeable and more annoying after normalizing than before. Basically, I get the vocal as clean and polished as I can before:

5) Add the music.

I've said this before, but the music I use (chosen for the cheesy 1940's 'educational film' quality and the fact it's public domain), is only about two and a half minutes long. I used to start with the original file, cut and paste to lengthen it and then go through the whole podcast adjusting its volume so it doesn't get distracting or inaudible at places.

That's the other drawback of this music. It has some nice loud sections and other very soft sections. I have to adjust the volume as it goes along, otherwise we end up with parts drowning out the vocal and other parts where you can't hear it at all.

Eventually, I had a 'duh!' moment, when I cut and paste the audio to make it about 10 minutes long and adjusted the dynamic volume for it. That way, I just drop this file into the timeline and cut it to size (using the quiet parts so you don't notice the join).

6) Listen through.

Inside Audition, I listen to the podcast all the way through, making sure I didn't leave in a mistake or accidentally cut out part of the script, and to make sure the music levels are working.

7) Downmix and upload.

This time, I mix down the music and vocal to a high-quality, but mono mp3 file. Then I go ahead and upload it to my hosting service.


So, long story short, you can't make a silk purse out of a pig's ear. While the sound quality basically sucks, it's the best I can do with the equipment I have.