Adobe Audition Podcast Tutorial
for speech recordings
This tutorial was originally written in 1998 as a tutorial for creating streaming audio (aka podcasts) using CoolEdit 96
Since 2003, Adobe bought out Syntrillium and CoolEdit Pro became Adobe Audition
CoolEdit 2000 later became Adobe Soundbooth
This basic tutorial will show you how to use Adobe Audition (or any similiar audio editing program) to record,
edit and compress a sound file suitable for making available on your site for streaming audio.
1) Get a sound editing program
Since Adobe Audition
are fairly expensive (about US$350 or US$200 respectively),
you can use any decent audio editor you have access to. One free
alternative is Audiacity
an open-source product, which as a bonus is available for Windows, Mac and Linux. The screenshots and the
step-by-step directions won't match exactly if you use a program other than Audition, but the basic principles
contained herein apply no matter what software you use. For a specific tutorial on Audacity, see
2) Start recording
I'll assume that you've got your audio source (tape deck, MiniDisc, mixing board, etc) properly connected
to the line-in on your soundcard and got the Windows Mixer set to record off line in (your soundcard
documentation should cover these steps if you're not sure how). In brief (assuming Windows XP or similar):
- Physically plug your audio device into the "Line-In" jack on your soundcard
- Go to Control Panel -> Sound and Audio Devices -> Device Volume|Advanced
- Change to recording level mode by going to Options -> Properties -> Adjust volume for|Recording (and make sure all boxes below are checked), then click OK
- Check the "[x] Select" box under the appropriate source, which could be either "Line In", "Analog Mix" or "What U Hear".
- Make sure the volume slider is high enough for the selected source. Usually set at 100% is good, and then adjust the recording level with your playback device. However, if you get distortion and/or you have you set your device playback level below 20% volume, adjust this slider so that playback volume and recording volume are both around 50%, and you're not getting clipping or distortion in your recordings
Start up Adobe Audition, then click on the Record button (red dot in the toolbar area in the lower-left corner
of the screen). You will be presented with a dialog to select channels and frequency.
Always use 16-bit, never use 8-bit or 32-bit audio format (8-bit sounds very bad and higher than 16-bit is not well supported).
You want to capture the optimal sound quality during the recording phase and worry about compressing the
sound later. Frequencies above 44100Hz are not supported on all sound cards, so stick with 44100. If you have
a stereo source (for music only, stereo for speech is completely unneccesary) you should select "stereo", otherwise
If you've selected Timed Record (see Tips
section below), a dialog box will appear
asking you to enter the time in minutes and seconds you want recorded (this example shows 45 minutes). It doesn't hurt
to record a little more just in case your time estimate is slightly off - it's easy to trim excess. If you have not
enabled Timed Record, skip to 2c)
. Note: Make sure Start Recording is set to "Right Away".
Recording is now in progress, and will continue until you either click the Stop button (or hit the Escape key)
or until the specified time has elapsed if you have Timed Record enabled.
Keep an eye on the volume meter (make sure "Show Levels on Play and Record" is checked under the Options menu)
and make sure the level never
goes above -1dB at any time. It's far better to keep the recording volume
a little on the low side (with peaks around -6dB) and amplify it later rather than record at too high a level
because once the audio is clipped it's pretty much impossible to make it sound good again.
3) Edit the recording
You should now have a recording that looks something like this picture. If you haven't worked with
an audio-editing application such as Audition before, left to right represents time, and up and down represents volume.
This picture shows a sample recording as it might appear before you do any editing. You can see that the
recording is not as loud as it could be (it doesn't reach all the way to the top and bottom of the screen). The
first edit we will do is to "normalize" (amplify as much as possible without clipping). To do that, click
on the Transform
menu, then under the Amplitude
submenu, click on Normalize...
Set the options in the following dialog to match this screenshot (the Normalize Left/Right Equally
is disabled if you have a mono recording). I suggest setting to 98% rather than 100% to avoid possible
clipping in the decoding
phase of the final compressed file.
The file should now look something like this, with the loudest portions reaching all the way to
the top and/or bottom of the screen. The next step we will do will make the audio easier to listen to for
the average computer's speaker design, especially for speech.
You can see that the loudest parts of the
recording reach maximum volume, but most of the time the average volume is much closer to -5dB, -10dB, or
even lower. To make speech easiest to understand, it's best to have the volume of all parts as close to the
same as possible. To do this, we'll make use of the Dynamics Processing
feature in Audition, which you
can find under the Transform
menu, then under the Amplitude
submenu, click on Dynamics Processing...
In the dialog that comes up, you will have to create the profile yourself, there is no preset that exactly
matches what we want to achieve here. First, make sure the Splines
checkbox is checked, then drag the
square from the upper-right corner of the grid down a little to -10dB. Then click in the grid where you see
the other two white boxes in this picture, at approximately -20dB/-10dB and -30dB/-30dB. Alternately, it might
be easier to enter numbers on the Traditional
tab of the dialog, as shown here. Select Compress 50
dB, Expand 2
:1 Below -20dB, and Flat
dB. Make sure you enter
negative numbers for those dB thresholds. Enter 10
for Output Compensation. Click back over to the
tab to make sure the graph still looks like the shape you see illustrated here - if you mis-entered
some numbers the shape will be quite different and you should double-check your numbers. After you click OK and
the processing is complete, the file should look something like this, with almost all the volume at about the
same level, but not at maximum volume. Perform the same steps as we did early to normalize the volume to 98% again:
click on the Transform
menu, then under the Amplitude
submenu, click on Normalize...
settings in the Normalize dialog should not have changed since last time, but just for reference they should
be set to 98%
and DC Bias Adjust
checked and set to 0
The next thing to do is to remove unneccesary gaps of silence. All speakers will leave pauses of
varying length between sentences and ideas, or when turning to a Bible text, etc. When listening to the speech
live, in-person, or even on video, where you can see the presenter, you don't notice this so much, but when all
you have is an audio recording on a computer, especially one that's streaming over the internet, this may be
distracting. Not only that, but removing unneeded silence from within a recording will reduce the size of the
file you create, resulting in faster downloads for your listeners and marginally lower server costs for you.
To delete silence from your recording, select Delete Silence...
from the Edit
menu. In the dialog
that appears, set the numbers as illustrated, which differ somewhat from the default Adobe Audition settings:
Define "Silence" as below -38
dB for more than 1500
milliseconds, define "Audio" as above -34
for more than 100
milliseconds, and Limit Continuous Silence to 1000
milliseconds. Once you have
entered those numbers, click the "Scan for Silence Now" button, and Audition will scan through your file and
pick out any sections that have more than 1.5 seconds of continuous silence and delete all the silence after the
first 1.0 seconds. You will usually find between 1 and 3 minutes of silence to delete in a 30-minute speech
recording. Click "OK" and the silence will be removed from your recording.
The final editing to do is trimming the start and end of the recording to make sure that the recording
begins and ends where you want it to. Highlight a small portion of the beginning of the recording with your mouse,
then click the "Zoom To Selection" button near the bottom-centre of the window. Once you're zoomed in, you can
highlight the exact part that precedes where you want the recording to start (use the Zoom In and/or Zoom Out
buttons as neccesary to view as much of the recording as you need to). Once the part you don't want is highlighted,
hit the Delete key on your keyboard (or choose Delete Selection
from the Edit
menu). Once you have
trimmed the beginning, use the Zoom Out Full button to view the entire recording, and repeat the process to trim
unwanted recording from the end.
This is the end of the editing section.
4) Save & Compress
Save your recording in the standard uncompressed WAV format. Select Save As...
from the File
menu, type a filename, navigate to where you want to save the file, and select Windows PCM (*.wav)
from the drop-down list of file formats. Click "OK" and your file will be saved.
The WAV file format is uncompressed - it takes up a lot of space, but there is no loss of quality. In the
next step we will take this large file (one hour of mono audio will be around 250MB) and compress it down
to a much smaller file suitable for distribution over the internet. The method I will describe here will
allow you to compress to either MP3 (probably the most common compressed audio format in existance right
now) or Ogg Vorbis
, a newer, free and
better-quality-than-MP3 compression method. There are other compression standards out there, including
, but while they generally
offer good speech compression, they are proprietary formats and limited availability of playback software,
with no guarantee of future support (or even the ability to convert out of the format to something else) if
the company developing it decides to stop doing so. That being said, if you do want to compress to a format
other than MP3 or Ogg Vorbis, you may go ahead and do so from the file you just saved and skip the next section.
Compressing audio to really small file sizes is a somewhat tricky to get optimal sound quality
from an appropriately small file, so I wrote this small program to automatically pick the best settings
for each encoder, depending on the type of content you are encoding (music or speech, mono or stereo).
First, you will need to download the software:
is an open-source MP3 encoder that is continually being improved
for more info on LAME).
, specifically the OggEnc.exe
file, is the Ogg Vorbis encoder.
Once you have downloaded and saved (and unzipped if neccesary) the files, run AudioArchiveCompressor.exe
The first time you run it, you will need to specify the location for OggEnc.exe
otherwise you won't be able to encode anything. After that, you can open the WAV file you saved earlier (drag & drop
works too). It's optional, but highly recommended, that you enter the title, speaker/artist, date, and Scripture
Reading (if appropriate) - this information is stored in the ID3 tag (for MP3s, Vorbis comment tag for Ogg). Track
number should be used to identify the order in which multiple recordings on the same day take place (set to "1" if
there was only one recording on that date). Choose which format you want to compress to (Ogg or MP3), and what
Genre (speech or various types of music), and whether it's mono or stereo (only mono is allowed for speech recordings).
Once the settings are correct, click the Compress button and another window will open showing the progress of the
compressor. The file will be compressed to a file of the same name as the source WAV file, but with a .mp3 or .ogg
extension in place of the original .wav extension.
You're now done! You may now upload the file to your website.
Right-click on the dB meter at the bottom of the window and choose 60dB Range rather than the
default 120dB Range - this will more closely relate to what your ears will perceive as volume.
You can also right-click on the right-side axis and choose to display in decibel format, probably the most
useful of the 4 available scales, since dB is a logarithmic scale and somewhat hard to visualize if you're
not used to it (and the other scales are linear (non-logarithmic) which might be confusing.
If you're recording off a previously-recorded source (audio cassette, MiniDisc, etc - as opposed to recording
the live performance), and you know the length of the material you wish to record, you can turn on Timed Record
from the Options menu - this will allow you to let the recording go unattended, it will stop when however many
minutes and seconds you specify have been recorded.