I’m a big Adobe Audition fan, and when I heard that the powerful multitrack audio editing program found in Adobe Production Studio would be replaced in the upcoming CS3 bundle by a more feature-limited version, I was less than pleased. That said, calling Soundbooth, the program that replaced Audition, "video-centric" rather than "feature-limited" is not smooth marketing talk. Soundbooth includes most features that video editors will need and is easier to learn than Audition. It also boasts a new background music creation feature that has promise, if not polish.
In this tutorial, I’ll provide an overview of Soundbooth’s interface and then focus on how to accomplish common tasks like normalization, noise removal, and audio compression. For the record, I created this tutorial on a quad-core Mac Pro with two dual-core Intel Xeon Woodcrest processors running at 3.0GHz and 8GB RAM. Not that it matters, of course, because Soundbooth looks and feels identical on both platforms, but it’s hard not to feel a small thrill running Production Studio on a Mac for the first time.
Soundbooth’s interface is similar to other Production Studio apps, using multiple interlocked panels that you can arrange as desired, then save as a workspace. You can import a video file into Soundbooth and edit the audio there, which is sometimes useful, though most users will pull their audio from Premiere Pro. Note the middle panel on the left in the image above, which contains tabs called Tasks, Effects, and Markers. This breaks the Soundbooth workflow into three distinct steps; we’ll work through the first two here.
The Tasks tab relates to cleaning up your audio, removing transient noises, and background hiss and hum. Soundbooth offers a useful selection of 11 effects; I’ll use a Compressor effect to boost the clarity of the interview I’m editing. The third step, the Markers tab, involves inserting markers for use with Flash, which goes beyond the scope of this tutorial.
Below that is the History panel, which stores your edits and allows you to undo them. Note that you can’t pick and choose which edits to undo. For example, if you undo the sixth edit out of nine, you also undo seven through nine.
Working in the Record Window
Though I’ve produced a lot of narration in Audition, the recording interface always felt like an afterthought. Adobe fixed that with a separate Record window—a nice addition (left). As before, when you’re working with audio in Premiere Pro, you can right-click and choose Edit In Adobe Soundbooth, and any changes will automatically be saved to Premiere Pro.
The biggest feature found in Audition but missing from Soundbooth is the multitrack interface, which is great for assembling multiple audio files into a single mix. For example, I use this to create radio advertisements for my wife’s ballet company, with one track containing voice-over and the other containing background music. Though I could easily do this in Premiere Pro, since there’s no video, and because I created the narration in Audition, I produced the audio there as well.
From a timing perspective, it’s interesting that Sony added multitrack capability to the latest rev of Sound Forge and Apple extended Soundtrack Pro’s existing multitrack capabilities with multichannel support. Note that Audition is still for sale as a standalone product for those that need that functionality, though it’s only available on the Windows platform. (Soundbooth, like most of the other applications in CS3 Production Premium, is available for both Windows and Mac.)
With this as background, let’s work through the most common tasks that you’ll perform in Soundbooth. I’ll be working with several clips pulled in from Premiere Pro that have random clicks as well as background noise that I’d like to remove.
Removing Random Noises
I’ve had to remove my share of transient noises in audio, such as random clicks and pops from a microphone. Or a high-powered rifle that interrupted an outdoor wedding ceremony last summer. Six times. By the sixth shot, the groom was really sweating, even though they were about a half-mile off. We don’t take terms like "shotgun wedding" lightly here in Galax.
Soundbooth offers an automatic tool that will scan your file for clicks and pops (left), but this didn’t work well in my tests. Rather, I had to select the transient noise and then use Soundbooth’s Auto Heal tool to remove it. In Audition, Auto Heal is called Remove Transient, and it’s a remarkable feature that replaces the noise with ambient sounds surrounding the undesired noise. It ends up sounding totally natural.
With Soundbooth, you can view the file in waveform and spectral view at the same time (left). I’ll assume that you’re familiar with waveform; the spectral view displays the audio by its frequency components. The dual view is great because because some transient noises—like high-powered rifles—show up more clearly on the waveform than in the spectral view. Other noises, like the cell phone ring highlighted in the bottom window in the image on the left, are indistinguishable on the waveform but easy to spot and eliminate in the spectral view.
Soundbooth supplies several selection tools to help you carve out the offending noise, as you can in the Remove a Sound box on the left of the image. The traditional time-selection tool on the extreme left works well for the waveform view, while the selected marquee tool, which I’ve used in the figure, and the more free-form lasso tool on its right, work extremely well in the spectral view.
Interestingly, if you click the Play Selected Frequencies Only checkbox, the expected happens, and you hear only the highlighted noises. I say interestingly because the lack of a similar feature in the program’s noise reduction function is probably the most high-profile deficit in the program. More on this in the next section.
Once you’ve isolated the noise to remove, click Auto Heal and Soundbooth removes the noise. Between the tools and views, you should have no trouble identifying and removing any transient noises.
Removing Consistent Noise and Hums
Soundbooth’s Auto Heal is ideal for transient noises, but it does little for consistent intrusions like microphone hum, air conditioning, or similar noises. This is where you’ll use Soundbooth’s Noise filter. As with the Clicks and Pops tool, you can click the Clean Up Audio group, select Noise, and then OK and let Soundbooth identify the background noise itself. Or, you can highlight a short region that contains only the background noise—like the white stripe on the right side of the image on the left—and click Capture Noise Print.
Then, you click Noise once more and make sure that the Use Captured Noise Print checkbox is selected (as in the figure). Adjust the Reduction and Reduce By Sliders to produce the desired effect, and click Preview to fine-tune your settings, using the green button to the left to toggle the filter on and off. Typically, you’ll want to remove as much of the noise as possible without introducing distortion into the audio.
What’s interesting is that Adobe left out the Keep Only Noise preview option found in Audition (and Soundtrack Pro, and most other audio editors). This was a critical step in my typical noise-removal workflow, because it let me hear precisely the sounds that the filter was removing from the audio file.
When removing background noise from a narration, for example, you would click Keep Only Noise and know that if you didn’t hear the person speaking, you weren’t removing any speech-related sounds and wouldn’t distort the audio. On the other hand, if you did hear speech, you could dial down the filter settings until it was almost completely gone. I asked the Soundbooth product manager about why this option wasn’t available in Soundbooth and he responded that Adobe was trying to move toward a more automatic application of this filter. In truth, I produced very good results without this option, but it does eliminate a very valuable security blanket for those who have used programs that offer this preview function.
Adjusting Volume and Normalization
Maybe it’s just me, but with most videos that I shoot, I have to boost audio volume, if only a little. As you would expect, Soundbooth has several options for this. Most directly, you can boost the decibel level directly by selecting the entire file, or any region, then dragging the dB slider to the right (it’s now at +8.0 dB in the figure on the left). This boosts all sounds in the audio file upward by the selected amount. Soundbooth helps you avoid "clipping," or the distortion that can occur if you over-boost volume, by displaying the new peaks before you release your mouse button, which is the light green waveform in the figure (the original waveform is the brighter green region).
Or, to boost volume to the maximum amount possible without distortion, click the Louder button shown in Figure 6. This is called "normalization." To elaborate, normalization boosts the volume of all sounds in the audio file (soft and loud) equally to the maximum levels possible without creating distortion.
Note that normalization will have little or no effect on volume if the loudest regions in the file are at or near 0 dB, which is the loudest digital audio can get without clipping. For example, if the same clip contains the bride and groom’s soft vows that barely show as peaks in the waveform and wild raucous applause that approaches the top, normalizing the entire clip will have little impact. That’s because normalization is applied equally to all selected audio, and won’t push louder regions beyond 0 dB. In these instances, you may have to go region by region to normalize the audio file, selecting the valleys between the peaks and normalizing only those regions.
Again, the first time you click the Louder button, you normalize. The next time you click the Louder button, it boosts overall volume by 3 dB, but applies a hard "limiter" on the louder sounds to avoid distortion. This means that it boosts the softer sounds, but not the louder sounds, limiting the increase in volume to avoid distortion. Interestingly, this is a primitive form of "audio compression" that reduces the difference in volume level between the softest and loudest sounds (this, as distinguished from streaming compression, which I discuss below). Audio producers use audio compression to add punch to a speaker’s voice. For example, I often use audio compression with wedding vows to make the sound more audible. When I narrate training videos for web distribution, I always apply audio compression to make my voice stand out.
I characterized the audio compression performed by the Louder button as primitive, because when using this control, Soundbooth doesn’t customize the compression to highlight any particular type of sound. So if you need to boost dialogue, or perhaps music, to distinguish it from the background noise of a crowd, you might want to apply Soundbooth’s compression effect before maximimizing volume.
Applying Compression Our tasks are arguably a bit out of sequence here, because when I apply compression I typically do it before normalization or other amplification. But for discussion purposes, it’s better placed here. To apply Soundbooth’s Advanced Compressor, click Effects, then Advanced, then Compressor (Advanced). That inserts that effect into the Effects panel (left). Click the Effect Preset list box and choose the desired preset—I’m using the Voice Over preset.
You can sample your preset by clicking the Play button beneath the waveform, and you can toggle the effect on and off by clicking the green button to the left of the Compressor (Advanced) effect. If the sound isn’t quite right, click Settings to open the window shown in the figure on the left.
For most users, these controls are so technical that you’re better off choosing a different preset rather than trying to adjust these to your liking. However, if you’re an experienced audio editor, have a go. Note that Soundbooth also offers a Vocal Enhancer (found under the Effects tab) that you can customize for male or female voice and music. If you don’t get good results using the Compressor effect, give that one a try.
Once you perfect your settings, click the Apply to File button on the bottom of the Effects panel to apply them. Though the effects will disappear from the panel after you apply them, you can always undo these effects within the History panel. If you’re producing a podcast or streaming video with the audio you’re editing and plan to apply MP3 or another form of streaming compression to your audio, you should render your audio to your final format before finalizing your compression settings. That is, if you’re creating a podcast, adjust compression (and then normalization) to the desired levels, then produce an MP3 file to test the result. If you plan to stream using the Windows Media codecs, encode your audio/video file to your final Windows Media parameters.
That’s because the interplay between audio compression and streaming compression can often produce undesirable results. For example, with male speech, you can use audio compression to deepen the voice a bit. Sometimes streaming compression can exacerbate this into harshness. So render your audio file to your final output settings before finalizing your compression settings.
This brings up a critical point on saving intermediate files while editing. Though you can undo all of the tasks described above while you’re editing the file, once you save the file, you can’t go back—Soundbooth applies all edits and updates the file. There is no concept of a Soundbooth "project" where you go back and change your parameters. So, if the noise removal filter produced some distortion that you didn’t catch during editing, and you’d like to try different parameters, you’re back at square one, which is always frustrating.
To avoid this, save the original audio file before any editing, with some sort of descriptive name (name–original.wav), then another file after removing transient noises (name–remove pops and clicks.wav). Then save another after noise removal. Note that Soundbooth keeps your original file open (and editable) while you save the others.
When you complete your editing on the original file, save that file. If it originated in Premiere Pro, the audio track there will automatically update. If you’re producing an MP3 file for a podcast, you can Save As, then choose MP3, and Soundbooth will open simple MP3 compression controls (left).
If you’re editing a video file in Soundbooth, you can open a full version of the Adobe Media Encoder and produce the file in any supported format, which includes Flash, QuickTime, and Windows Media.
That’s it. Though Soundbooth isn’t quite as cushy as Audition, it handles most critical tasks with aplomb, on both the Mac and Windows platform.
And unlike Audition, it was designed specifically with video editors in mind by a company with a wealth of experience in identifying video editors' needs and developing its products to meet them.
Jan Ozer is a contributing editor to EventDV and Streaming Media and the author of Adobe Digital Video How-To's: 100 Essential Techniques with Adobe Production Studio, published by Peachpit Press.