What is Hiss?

Not to be confused with hum, hiss is a persistant noise on a recording that sounds like … well … hiss. Turn the sound up loud enough and you will almost always here some on your recording. A low levels, it isn’t much of a problem.

The problem is when you have hiss on a recording that is distracting – say from a bad microphone or from turning up low dialogue. You could also be trying to restore an old recording that has hiss on it. Wherever it comes from, however, it’s annoying and needs to be destroyed.

The Basic Method of Getting Rid of Hiss

Hiss removal involves a few similar steps no matter what software you are using. First of all, you need to find a sample of hiss just by itself. That way the audio program knows what the hiss actually sounds like isolated – before that, it could get hiss confused with other parts of your recording. After setting some parameters, your tool of choice will try to remove that hiss from the recording without damaging any other areas.

To illustrate, I am going to take a look at removing hiss from two programs: Audacity and iZotope RX.

Removing Hiss with Audacity

If you are not familiar, Audacity is a great (and free) audio tool for Mac and PC. You can get it here. To get started, open it up, and import a piece of audio.

After importing your track, look closely for a small chunk of audio where there is only noise. After you’ve found it, select that portion of audio.

Select Noise Removal… out of the drop down list of effects from the Effects menu. This will give you a little dialog box that presents a handy two step guide. The first, is to get a noise profile. Since we’ve selected our noise profile area, we can just click the “Get Noise Profile” button.

Once you have the noise profile, deselect the noise profile area, and get back to the two step panel. You can adjust the noise reduction amount, preview it, and hit okay. Play back your audio, and the hiss is now minimized or gone.

As we mention in our limitations section, your audio may be affected depending on the Noise Reduction level you’ve chosen. If you are getting audio that sounds like it was recorded underwater, it is time to go back and rethink your parameters.

Removing Hiss with iZotope RX

iZotope RX is a fantastic program that is well worth the fact that it is not free. You can get it at BHPhoto.com.

What does iZotope RX have that free solutions don’t have? Well first of all, it can do a lot more than take out hiss, but we are sticking to hiss in this article. One of my favorite features of iZotope RX is its Spectrogram information. It can be viewed right on top of the more conventional wave pattern:

This view allows you to see – in much greater detail – disturbances or irregularity in audio. For instance, hum can be observed as a single line shooting horizontally through a recording. Hiss appears as “noise” you can see in the background of the audio.

Removal works in a similar fashion to Audacity. Find a portion of the audio where just the hiss is present, click the Denoiser button and click Train. This will “train” the program in the audio noise profile of the recording.

As in Audacity, you have a Noise Reduction level and a Smoothing option. However, you also have a choice of several algorithms that might give different results. Some take longer to process, but it is worth it to see which one might fit your situation the best.

When using an advanced program like iZotope RX, it is a good idea to look through their documentation. In the case of iZotope RX, the documentation happens to be extremely helpful and thorough.

Limitations

Of course, there are limitations. Like we mentioned in our previous post, audio is like a bowl of soup. You are going to take out some things you want to leave in. Since every piece of audio is very different, you’ll need to do some experimenting with parameters to see what level of removal leaves your audio suitably in tact.

That wasn’t the case before, and still isn’t the case in many production environments. You see, more often than not, we modern editors are doing something without even noticing: we’re combining offline and online editing. Essentially, it used to be that you captured much smaller, lower quality versions of files, and edited those. That was offline. Then, you’d send your edit to another department (or do it yourself) and they would do the online editing, meaning they would put in the full quality files from the source tapes, color correct, etc.

Now, you probably capture at full HD and edit at full HD because our computers can handle it. It’s simpler, and for a smaller operation, it makes sense.

But even though we’re editing in different ways, the fundamentals under the hood are still the same.

Numbers Under the Hood

For this example, I’m going to use Final Cut Pro.

Let’s say you have a 45 minute tape of HD footage, and you capture it into FCP, end to end. Now, that tape is file on your hard drive, sitting in a folder. Go to your project assets folder, and you can check it out and even play it from the finder.

However, go into Final Cut, and you can chop it up, move it around, color correct it, and mess with the audio. After all that, you can go back and still find your 45 minute file, just sitting there, not touched.

The reason for this is Final Cut is a non-destructive editor. It’s not actually touching your captured files at all, it’s just referencing them and creating data about the different editing decisions you’ve made. A cut on this tape at this time, placed on the timeline here, with this filter and this data. That’s all it needs to know, really, and then it plays it back.

When you get down to it, editing is all about numbers: timecode and other data referencing tapes. Those tapes just happen to be sitting in digital form on your hard drive. That’s why your Final Cut Pro project file is always so small. It’s just data – the files are elsewhere.

Could I Get That In a List?

It makes sense then, that since you can export a video file from Final Cut (where it pulls together all that data and mixes the video together into a file), you can export a version of just the data. This is called an Edit Decision List, or EDL.

EDLs are a list of every editing decision for an entire project, in sequence. Theoretically, if you lose your entire project, if you have the EDL, you can rebuild the edit. Pretty crazy.

Now, like many things on the computer, there are many types of formats that all vary slightly (thanks, computers), but most EDLs share the same basic data from the classic CMX EDL format:

Event Number:  Just a sequential number. Primary key, for my database people.

Source ID: ID for the video or media source. Some programs place limits on this in terms of length.

Edit Mode: Is this decision on video track 1 (V1) or audio track 1 (A1) or both (VA1) or maybe another combo? That goes here, and the format in which it is denoted varies.

Transition Type: C for Cut, D for Dissolve. You get the idea, and you also get the frame numbers in there for dissolves. These also vary and are not standardized.

Source in/Source Out: Timecode of where you are taking the video from the original source.

Record in/Record Out: Timecode of where the video you took appears on the tape or timeline.

For example, here is a CMX-format EDL for a very short, basic video with straight cuts:

You can also get a much more complicated version:

As you can see, in Final Cut, there are a lot of EDL options to choose from. Which ones you use (and which ones you use on any program) depends on the needs of the project. The person who needs the EDL will usually be able to tell you what kind of settings they need on the export. What you need to know is that EDLs exist, and in many cases are essential to the production workflow.

So, let’s talk about codecs. You know you want to.

Let’s say you’ve got a full HD video of something. Maybe it’s 10 minutes and it’s like 3GB. Now you want to make a version that is 3MB so you can send it around to a couple of your friends to get some notes on the edit. Maybe you run it through Compressor or FCP or whatever. You get it down to its desired size, and you’re good to go.

Somewhere along the way, you choose which codec you want to use, and you may just randomly select something or something that sounds familiar. Now you’re on your way.

It may never cross your mind, but how exactly does your video get from 3GB to 3MB? It loses some quality, yes, but why and where? What’s the logic?

The logic behind how your video goes from big to small or small to big depends on the codec used. Codecs basically store the logic behind how to compress a video and they contain the logic to compress and decompress video files, hence the term codec. It’s a standardized method.

You might be asking “is there just one codec that does all the work for all video?” Well, no. There are many, many codecs. And you’ve probably heard of many of them. Here’s some codec name dropping: VC-1, H.264, MPEG-4. Sound familiar? Those are some of the big players, but there are hundreds of codecs out there.

There are an infinite number of ways to compress or decompress something. You can do it lossy or lossless. Many work to find patches of similar data and store those in shorthand. Some simplify color. The common thread, though, is as technology advances, the algorithms get smarter, and we get smaller file sizes with higher quality video.

By the way, just like pretty much everything in technology, companies create codecs and license them to make some cash, but many are open source.

The important thing to remember about codecs is that they don’t just exist on your computer. Your BluRay player has one (capable of playing back H.264 and VC-1, among others.) And guess what? Your camera also uses a codec. My HVR-A1u says it uses the HD Codec Engine ™. They’re everywhere video goes through a pipe, pretty much. From lens to tape, or lens to hard drive, and from disc to TV set.

To learn more, check out Wikipedia’s list of codecs (scroll down for video codecs). They are way too many, and they are too different to get into here, but in the future we’ll do a primer on different codecs and what they are best suited to.

Got something strange or belligerent to say about codecs? Do you hate codecs? Spit it in the comments!