[Unit 55]

1.1

Editing audio is the process of taking what you've recorded either in the studio or in the field and using a computer to edit these files. This edit will attempt to bring out the best in the audio either by removing background noise or enhancing vocals.
Many programs will automatically translate an audio file into visuals that represent the sounds in the file. When we listen to audio our brain is interpreting the vibrations in the air that enter our ears into electrical signals that we recognise as sound. These vibrations can translated into a visual form. This form can either be a waveform that shows amplitude over time or a spectral graph which shows different intensities of colour for different frequencies.

Green section is the amplitude waveform measured in decibels up to 0db
Purple/red section is the spectral graph which is plotted against Hz (purple is less intense than yellow)

Most people have seen a vinyl record and a CD. These two mediums have different types of audio on them. Vinyl uses an analogue signal while a CD uses a digital signal which is comprised of 1s and 0s. It is argued that analogue is a much fuller sound but it can also have it's drawbacks because of hiss or cracks from the imperfections of the vinyl press. Digital sound is said to be of a lesser quality than analogue because naturally it has less data but obviously it is more versatile because it can be transferred and copied infinitely. You can see from the picture below the loss of data represented by the white areas between the black analogue signal and the digital bar signal.

y-axis measured in sample rate

 x-axis measures time.

There are many things you can do when editing digital audio. This is another reason for it's great versatility. Using programs such as Audacity or Adobe Audition enables you to copy, cut, paste, silence and add effects to an audio track. You can do this using the different visual representations of the audio. The spectral graph is especially useful because because it gives you great control for example, being able to delete specific frequencies in an audio file. Editing software can also allow you to layer different sounds. You may use separate audio tracks to work together to make a new overall sound. For example you might use a voice over in a documentary but also have some ambiance in the form of quiet music. 

• Importing audio files in the audio editing software is fairly easy. You can record directly into a lot of audio software whether it be from a microphone or musical instrument. This can then be edited in ways previously mentioned. 

Importing is as easy as dragging and dropping or choosing import
and then selecting the file you want

• Most audio editing suites offer many different editing features other than just cutting and exporting. Cross-fades between clips, silencing certain points, using the heal tool to correct unwanted sounds. You can also use a brush too to make specific selections of audio that you can then use effects to manipulate.

Here you can see the while line of the heal tool. It is
being used over an unwanted sound in hopes that it
will be removed and the audio will sound clearer.

• For recording voice there are a few different effects that can be applied to improve the clarity of the audio. One effect you might at is the parametric filter. This effect allows you to increase the amplitude of certain frequencies of your file. If something is too bassy you can either reduce that bass or increase the higher frequencies to add some more treble and 'crispness' to the voice over. 

Here you can see the multiband equaliser. This can be used to enhance vocals by
bringing up the middle frequencies and reducing the high frequencies to give a 'fuller'
sound

• With most audio files you want to have as high a quality file as possible. There are usually many different formats that you can export an audio project as. If you want a lossless format then .WAV is a good choice. This is uncompressed but as a result has a large file size. If file size is an issue then you may use .mp3 which is a good compromise between quality and sound and if you choose 320kbps then it is often very difficult to distinguish between an lossless format and this.

Here we are selecting the export options for the finished file.
The original was recorded at 48000hz in 24bit so it makes sense
to try and retain this quality as much as possible.

1.2

There are two different type of editing processes when editing audio files, destructive and non-destructive. They speak for themselves but to put into words, destructive is where you edit and make changes and whatever changes you do make affect the master file you're accessing. You can however get around this by making a copy of the file just for editing. Non-destructive is a process where you don't affect the master file. With this process you can layer different audio tracks and apply effects without actually finalising them.

1.3

A recent brief we were given by a client was to record audio for a voice over on a presentation. We were to record someone giving the presentation which was supposed to last all day (9am-5pm). Because of this the time that we needed to allocate for filming and recording audio was a full day. We would also need a two man crew to operate two cameras and the audio equipment. Post production would also need a full day for syncing audio, editing clips and outputting the final videos ready to be sent. I'm not directly involved with the budget for certain shoots but all the equipment we used was our own so there was no additional cost to the client other than the rate for a two man crew for a full day of shooting + one day in post production.

2.1

Carrying on from the previous question which mentions a shoot we went on the record a voice over, the brief also included what the final product should be like. The voice over is intended to overlay a PowerPoint style presentation video. It included 9 sections with of varying lengths and number of slides. The voice over is to coincide with these sections to articulate extra information to the viewer. The intention of the product is to be informative and because of this must be clear, understandable and not too convoluted. There was no need for additional sound effects or audio because the voice over is all that was required. This means there was no need to use any multitrack editing and in theory all the editing could have been destructively (while still keeping an unedited master file of course). It was important to make sure that there was minimal reverb and floor noise so the presenters voice was clear and understandable. 

2.2

The final produced material will have a final destination on the client's website. Because the audio will be used on top of a video and the file format we use when providing material to clients is .MOV, the audio format was AAC at 320kbps. The sample rate was also 44.1kHz to match the original file. This provides some compression to keep files sizes down but is also a good compromise on quality. Using a lossless codec would have greatly increased file size meaning it would take longer to output and upload to the client. The original audio was recorded in mono so the final output was mono also. There wasn't any particular reason to use stereo as we were only recording one subject.

2.3

There a many programs both free and paid for that can be used for audio editing. A popular free program to use is Audacity which has quite a few editing tools and effects that can be applied. You can also record straight into it from an external mic or instrument which is a useful feature for beginners getting into audio editing. A paid alternative could be something like Adobe Audition which is part of the Creative Cloud package. Being a paid program means there are quite a few more effects and tools that can be used to manipulate audio tracks. You can edit destructively in Audition by directly altering the file or you can use the multitrack feature which is not destructive editing. You can also layer different sounds using this method which can be especially useful if you're creating a more complicated audio file which needs more then one voice over or sound effect to be present. For example you may want to put several different instruments on top of one another to produce a song. Both programs have basic tools like the blade tool and changing the gain. I know that in Audition you have tools/effects which allow you to de-hiss, de-ess and de-clip which are all very useful for producing good, clean audio. There's also more involved tool like the parametric equaliser where you can increase the gain for specific frequencies in the audio to make to clearer or muddier. All this tools can be used together to create a high quality final piece.

Using the multitrack editor to layer audio files
(Audition)

DeClipper, DeEsser tools
(Audition)

Parametric Equaliser being used to bump lows and highs
to add clarity to voice over.
(Audition)

2.4

The voice over we had to produce had to be ordered to match the script we were given. In the screenshot below you can see folders including various sound files. Each folder corresponds to each script that we recorded. The files inside this folder are the audio that was recorded from our speaker. Each script was recorded 3 times so they're all roughly the same and this was done so that the cleanest parts of the audio could be spliced together to make a perfect end copy. At the very bottom you can see a folder called sound bites. Because most of the scripts included exactly the same paragraphs, we decided that we would record these sound bites so they could be dropped into their appropriate places on each of the audio clips. This saved the speaker's and our time as there was no unnecessary repetition. To be able to cut and splice these audio clips together, a multitrack project would need to be created which is what You can see in the screenshot to the right.

Audio files organised into appropriate folders

Multitrack creation in Adobe Audition

Here to the left you can see the imported files from the project folder ready to be placed in to the multitrack editor. This included the separate sound bites recorded and the audio for the main script. on the right you can see the audio files placed into the editor. I first listened to each of the three recordings and then used the razor tool to separate out the cleanest sounding audio. I rearranged these isolated parts and put them in the appropriate order. Any excessive gaps were cut out also so the audio flowed at a good pace.

Audio clips spliced and placed on timeline

Imported files for 'Counter assistant' script.

2.5

In addition to the voice over work we did for Buttercups, there was also an interview scenario that we were asked to film. I assisted on the audio recording for this. There were three camera angles we had to film and therefore 3 audio clips that needed to be recorded. The interview consisted of various pharmacy personnel being asked questions by an inspector. Each person had a lavalier mic which fed into a transmitter and into the DSLRs.

Each person also had a rough script that they were to stick to.

After recording took place we needed to organise the audio and cut it to make sense and flow nicely. The audio was extracted from the video clips using Audition. I took the time to fix any problems that may have been apparent such as hiss or unwanted clicks and noises. It was fairly simple to split the original recordings to and organise them because we had  rough script to follow. This script also meant we knew what message we needed to convey and there was no chance of getting any structuring wrong.

2.6

I haven't been involved with any instances of recording a music performance however I did recently visit Paper Stone recording studios in Sherwood to record a singing track for EON. There were four EON employees singing a song that had been made as a jingle to advertise smart energy meters. Our job was to record the whole recording session including the people singing and the coach training and giving encouragement to the singers. The brief for our job wasn't to produce a video of the singing but was more of a rushes package. Because we filmed the video separately from audio recording I had to match the audio with broll that was filmed by dave. Because the whole session was an hour and a half long, matching short broll clips of 8-9 seconds took a considerable amount of time. I did try to cut down some of the parts that didn't look or sound great but I kept in the more important sections of the recordings. Each verse of the song and the chorus were recorded several times so I included at least 2 of each of these so the client has a selection to chose from in their rushes package.

2.7

Determining your edit points is done by looking through your audio files and seeing where cuts and transitions need to be made. If you have a brief for what you are trying to achieve this will help you also. If you are following a script then you'll be able to see the points where you need to make a transition. When making these transitions you should try to be as smooth as possible. If you're cutting close the the end of words then the cut is going to sound harsh so it is always best to leave at least half a second between finishing a sentence and starting a new one. This gives the impression of natural speech which is what you want to achieve to give it a good pace and to flow nicely.

You can see in sections 3.1 and 3.3 where I have given an example of selecting in/out points and using the fade in-out to make the transitions smooth.

2.8

See 3.1 and 3.3 for example on trimming, length, gain and fading in and out as well as cleaning up background noise. Converting from stereo to mono can be done in the output stage which can be seen below in the screenshot

Here are the export options for a voice over I have edited for
the East Midlands Expo

3.1

There was a sound clip that I recently had to work on, to alter the sound and make it sound clearer. It took place in a workshop with lots of background noises and reverb due to the hard walls. The purpose of the recording was to inform via a voice over so it needed to be clear and intelligible. There are a few methods and effects that can be used to clean up some of the sound. You do however have to make a compromise because if you clean the audio up too much you get a very false sounding clip. Below are some screen shots of the steps I took to remove/reduce the background noise, to reduce the reverb and to increase the clarity in the presenters voice.

To rectify this you select a sample of the background noise
and then capture noise print

Here in the spectral view you can see the amount
of noise that is present without anyone even
speaking.

Here is the noise remover effect in Audition. Here
you can adjust how much the sound is reduced.
It can take a lot of trial and error to get it sounding
perfect.

After the effect has been applied you can see
here that the noise has significantly been reduced.

This is the parametric effect being applied across
the clip. Here you're able to increase the gain for
specific frequencies. In this case we lost of some
of the bass and treble of the speakers voice when
reducing the noise. Upping the gain at these low
and high frequencies can increase the clarity of
the voice

Here is the dynamics processing effect being applied.
This effect will reduce any reverb you may have picked
up during recording. It is an extremely sensitive effect
in the sense that adding too much can be very noticeable
in your recording

3.2

See 1.1 concerning amplitude and spectral displays.

In adobe audition there is very simple diagnostic tool that you can click which will analyse the audio. If you have a stereo clip it will analyse both tracks and give results for them both. In the screenshot below you can see the date is returns when analysing. It shows a stereo audio file but the results of the analysis is the same for both so it doesn't seem to be true stereo.

Analysis of the entire audio track showing max peak,
average dB and several other pieces of information

3.3, 3.4, 3.5

Like previously mentioned, the Buttercups project I worked on required me to blend different sound tracks so there were smooth transitions between the different sound bites. The brief we were given outlined we needed a clear and intelligible voice over for a presentation/slide show type film so I used certain effects to make the recording as high quality as possible. The intended artistic requirement is to convey to the viewer that this is a continuous one-take presentation. This means that I need to create seamless transitions between audio clips and the script should be error free to convey a quality presenter.

Here is the effects rack which is used to add multiple
effects to an audio clip. You can see that I've used
dynamics processing, parametric equalizer and a
vocal enhancer. See 3.1 for details on these effects.

Here you can see multiple tracks overlayed
one another. The yellow lines you see show the
fade in/out duration. This gives a clean transition
between the clips so there are no sharp
changes in sound or volume.

Here you can see the DeClicker effects. When using this you scan

your audio file and Adobe Audition will automatically

scan for any clicks. You can change the threshold to get rid of

less/more extreme sounding clicks

On this waveform view you can see the points in which the levels are different.
Having different volumes throughout your recording can be undesirable and can
be annoying/distracting to a listener. 

Selecting specific parts of the track where you can visibly see it's quieter you can
adjust the gain to bring the volume up to match the rest of the audio. This gives
you a more consistent sounding recording.

3.6

Sometime you may run into issues with compatibility when using different sound sources, especially when using different audio recorders and microphones. There was an issue recently when working on a project for MIRA that we had to use a voice over that was recorded by MIRA themselves that had a very different sound than what we recorded. The MIRA recording was done via an iPhone so the quality was much more tinny sounding than what we recorded with the RODE top mic. Using the effects I previously mentioned I attempted to match the sound the best I could so that it wasn't noticeable that it was recorded at a different time. The results turned out relatively well and was hardly noticeable. As for other problems such as background noise and clipping, there are simple tools that can be used in Adobe Audition that will get rid of these with ease. You can scan for clipping and repair these and with the noise reduction there is an effect you can apply that I have mentioned in another section. It is generally best to get the recording as clean as possible beforehand that way you spend less time in post production trying to sort it out. There may also be instances where different audio clips have different pitches and tones that may be distracting if they're constantly changing. You can use the pitch-shifter tool to change the pitch of certain sections without changing the speed. Keeping this consistent can make for a much cleaner sounding audio track and isn't distracting to a listener.

Here you can see the difference in background noise. This is also a good overall example of use of all the effects and tools that I have previously mentioned throughout the blog.




4.1 & 4.2

When we're using our external Tascam Recorder files are are automatically numbered consecutively each time a file is created. The general naming convention is 'TASxxyyzz000' e.g TAS041116001 showing the first recording on the 4th November 2016. We're also able to name folders on the Tascam but this is usually left until everything is backed up on the computer as it's much easier to change the names then. To get the highest quality we can we always record in WAV 24bit 48khz. This gives us a lot of flexibility in post. Files sizes may be larger but we use a 16gb SD card which is more than sufficient for a couple days of recording. There aren't really any guidelines we follow to document the material we produce. The only thing we do with our content is back it up to our central NAS. Each job we go on has it's own job number so when things are backed up our assets are put into a folder with the job number and a brief description.