Time Saving in Audacity

At the Complete Developer Podcast we’ve been using Audacity for our recording and editing. It is an open source audio editor with plenty of great tutorials. I’m learning to use it as we progress with the podcast. It is rather intuitive, though you have to know what you want to do or at least what the functions do to the sound. I had some fun playing around with the speed and pitch when first learning. Try dropping the pitch an octave then speeding it up 2x. It’s funny but not useful.

I edited and mastered episodes 0 and 1 in 30 hours over a weekend. That is approximately 15 hours each though more likely episode 1 took longer as it had more parts to edit and was a longer episode by several minutes. Much of this time was spent learning the program and abilities of Audacity. Also, our first few episodes were not all recorded in Audacity and therefore some had mono and others stereo tracks. When recording in Audacity we use the stereo tracks for better quality. This created extra work to get the sound quality to appear similar across the episode.

I was able to shave a couple hours off for episode 2 and finished it in 13 hours. While that is an improvement and doable at one episode per week it is still a lot of time spent editing. I came up with an idea which I’ve detailed below on how to save some time by combining my tracks early and separating our the various parts of editing. Well, if you saw my Facebook post the other day it worked! I cut my time down to just over 7 hours. What makes this more exciting is that the episode is our longest to date running a total of 42 minutes.

We record The Complete Developer Podcast in segments or modules so as to save time recording and in the editing process. In this post I’ll be working on the body of one of our podcasts since that is the longest module we record and it allows me time to write while I wait on Audacity. The other modules except the music are done the same but take much less time. This also allows us to record several segments of IoTease and Tricks of the Trade at once so in our limited recording time we are able to focus on the current weeks episode.

We store our audio files on a VPN to which we both have access. To begin I combine both tracks from each of our microphones into one new file I label with -mix. I also will label the individual tracks based on the mic it is from. The file names look like Episode3-social-body-mix.aud and an individual track will be labeled body-b.

I. Noise Reduction

To start off we use a tip we picked up from listening to Dot Net Rocks. This involves holding our breath for a few seconds at the beginning of the recording to get a sample of the background noise. We try to reduce this as much as possible before we start but AC and fans make a low rumble.

Since we’ve done this in Audacity I am able to select that section of the recording and then go to the Effects option on the top menu. Under Effects I select Noise Reduction and then at the top of the pop-up window select Get Noise Profile. This calculates the background noise to remove. I then select the entire track, you want to do each track individually since the noise profile will be different. Going back to the Effects option I again select Noise Reduction and then select Apply Noise Reduction. I repeat this on each track until there is significant reduction to not hear the background.

II. Isolate Vocals

After the noise has been sufficiently reduced I go back up to the Effects menu and select Reduce/Isolate Vocals. In the drop down I select the Isolate option and run this separately on each track. This helps to emphasize our voices of any residual background noise as well as any 7-year-old thumping around upstairs. This is a short process on my end but can take up to 10 minutes for me to wait on Audacity depending on the size of the recording.

III. Silence The Other Mic

Now comes the fun part and the little trick I learned that cut my time in half. In the first few episodes I was doing this while also editing for content and fine tuning. I thought that by doing it all at once I would save time. That was not the case.

This step involves listening to the episode and silencing the track of the person not speaking. I start this by lining up the tracks. There are some tools in Audible to assist with this but I have found that zooming in to a loud point that almost touches the edges I am able to do a better job “eye-balling” it and listening. It does sound funny when I get it wrong and we have an echo. Once the tracks are lined up I go up to the Tracks menu on the top and select Sync-Lock Tracks. This way I won’t accidentally move one and have to go through the whole process again.

Now I go through the episode and listen for my parts and Will’s parts. When one is talking I will highlight then silence the other track. It’s not that bad as there is a slight distinction in our voice patterns that I’ve began to recognize. Also Will’s speech appears larger on his track and the same for me. For the most part we have a good back and forth without much overlap but I have learned (if you listen you can hear where I didn’t do this in the first few episodes) that if we speak simultaneously to use the track from the person who continues after that. This helps with continuity and well it just sounds funny otherwise.

IV. Mix and Render to New Track

Here’s where I got smart, or less dumb. I then combine the two tracks into one. First I go to the Tracks menu and select Mix and Render to New Track. I do this so I’ll still have a copy of the original. I look over it for any obvious mistakes and listen to areas of uncertainty to make sure I didn’t miss an overlap or have the wrong track silences. It has happened and I was very glad I still had the two originals. At this point I should mention that on the left side of each track you have the option to mute the track or to make it solo (mute all other tracks).

If the mix is viable I save my work and label the mixed track body-mix. Well, I save throughout the process but things are about to change here. Then I delete the individual tracks and Save As Episode4-mental-full as I will be using this file for combining the full episode.

V. Normalize

I’ll do this a couple of times after mixing tracks. In the Effect menu is an option to Normalize. Using Normalize brings the mix to the same level. Right now it is less important but takes less time later if I do it on the large piece.

VI. Edit for Content

Things are now starting to get time-consuming. Here I go through the episode and remove any long pauses or big goofs that are obvious just looking at the track. Since I edit soon after we record I usually know where to look and where the conversation gets off topic. Will and I have been friends for about 15 years and we are both talkers so we can get pretty far off topic at times. A little trick I learned from a tutorial about removing segments is to highlight the segment to be removed then press “Z” to move the edges of the highlight to where the wave crosses zero. This avoids any pops when transitioning between the pre and post cut.

VII. Edit for Fine Tuning

Time for the fun part. Once again I will listen to the entire episode and look for pauses where we say “umm” or “like” or “right?” or make loud breathing sounds. I remove those as well as possible without making the speech sound unnatural. If you listen to the first episode it sounds like we are talking fast, well that was mostly me not knowing how to edit and leave enough pause for normal speech. I’ve found that if within a thought I can reduce the pause to 0.25 seconds and it still sound natural. Between thoughts or sentences I add 0.5 seconds if I remove an “umm” to make it sound natural. I add this time of silence by using the Generate menu and then selecting Silence.

VIII. Add All Individual Tracks Together

Once I’ve followed that pattern on all the modules of the episode I put them all in the -full file. In the Tracks menu is an option to Align Tracks. Under this is Align End to End. This is a great tool as I just align then sync the tracks as I add them to the body. After everything is added I’ll unsync the tracks so that I can adjust the music to fade in and out under the talking.

IX. Adjust Gain on Each Track

To create a uniform sound I go through and adjust the gain on each track. Also I have used the Envelope tool to lower some of the music so that it is not overbearing. In the first few episodes the IoTease theme jumped out at you when listening. From Episode 3 onward I’ve been able to reduce that so that it is the same volume as the rest of the episode. I know that if it annoyed me it must have annoyed a few listeners.

X. Mix and Render to New Track

This is the same as before. I use the Tracks menu and select Mix and Render to a New Track. I then check the new track, specifically the transitions with music involved. These are the ones that have caused the most trouble. Once the mixed track is approved I label it -full and save. I then delete the other tracks as they are all saved separately. I know I’m being redundant but I don’t want to have to record again and definitely don’t want to have to tell Will I lost the whole episode.

XI. Normalize

Again I will go up to the Effect menu and select Normalize. This time mainly to make sure the music sections are not overbearing the speaking.

XII. Review for Errors

This is important. I’ve finished the mixing of the full episode. At this point I will usually take a break, especially if I’ve been working for a while. I like to go watch an episode of Doctor Who or play some 360. Just to get away from the screen for a while. I’m sort of trading one screen for another though. I should take my dog for a walk during this time, he would love that. When I get back I review the episode. I won’t listen to it entirely here but will hit the transitions and a few random spots.

XIII. Set Meta Data

Once I feel satisfied or exhausted, you’ll be able to tell which when you listen, I start adding the meta data. I like to do this before exporting to mp3 format so that I have it when I save the Audacity file. In this I include Artist Name, Track Name, Track Number, and Album Name. The iTunes image is added later when we publish it to our CDN. This is done automatically since it is set for the Album and that is the same across episodes (Complete Developer Podcast).

XIV. Export to mp3

The final step is to go to the File menu and select Export Audio. I then select the format (mp3) and click OK. It takes a few minutes then I save it to our server on the VPN and to Dropbox as a backup. Finally I upload it to the unpublished media section of our CDN (Blubrry).