30.5.2022 - 18.6.2022 (Week 9 - Week 11)
Chung Yi Ki / 0345014 / BDCM
Sonic
Design
Task 3: Audio Storytelling
Lecture
Week 9 / Noise reduction
Removing noise using the traditional method
- Select the part with noise and then go to effects > noise reduction > capture noise print.
- Then select all and go to effects > noise reduction > noise reduction process.
- In the noise reduction process, the noise can reduce by controlling the two parameters and by controlling the frequency range in the equalizer-like window.
- In the adaptive noise reduction window, noise floor is the noise level in the volume bar. If the voice starts to sound weird, then the threshold might need to be adjusted.
Cleaning up the vocal track
The volume of portion of the sound can be adjusted by selecting the area and
reducing the volume, this is how spikes in recorded audio can be pushed
down. Another way to control spikes is in the effects, go to amplitude and
dynamics.
In the dynamics window, under compressor, the threshold means anything out
of that range will be pushed down, the ratio means for every dB reduce by
given amount. Attack means how fast you want to compress the volume and
release is how fast you want to bring the volume back up again, usually a
shorter timing is better in a normal speed recording. The make up is to
increase the volume of the whole audio since compressor will lower it (like
input boost in hard limiter). But before using compressor, the audio needs
to be manipulated manually, because if you didn't then the loudest spike
will be used by the compressor to reference and this will affect the other
spikes.
Any breathing, pops and clicks are reduced after the noise reduction and
compressing the audio, this can be done manually by selecting the area and
decreasing the volume. An automation way of doing this is to use Autogate in
the dynamics window. Threshold controls the decibel value to be silenced,
anything out of that volume will be heard.
Final touch ups
After the clean up process, use the hard limited to finalise the audio.
Adjust the input boost so there's some waves in the audio, not just boxy
visual representation. Be careful not to make anything boxy, the aim is to
make the voice consistent from one sentence to another and not just make all
sound waveform boxy.
Tasks
For Task 3, we are required to either choose a popular fairytale story or
write one ourselves, and record narrative and voice acting for it. The
recording will later need to be edited in Adobe audition to make it sound
clearer and better, as well as adding any suitable sound effects and music.
Then, the final exported audio needs to be paired up with either screenshots
of the storybook pages or screenshots from a movie/show/illustrations. The
duration of the video shouldn't be more than 3 minutes and the narration and
voice acting can be spoken by a friend. It is preferrable to record the voice
using a mobile device.
Story selection
Story selection #1
A good kitty and a bad egg (2011)
Link: https://archive.org/details/goodkittyandbade00oliv/mode/2up
(sign up is required)
Story selection #2
Disney's Magical World of Reading: Robin Hood (2007)
Link: https://archive.org/details/disneysmagicalwo00dal_jsf/page/n1/mode/2up
(sign up is required)
Story selection #3
Link: Disney's Aladdin Mini Storybook (2009)
Link: https://archive.org/details/aladdin0000unse_e6k9/mode/2up (sign up is required)
My first thought was to find any storytelling-like summary for the movie "Puss
in Boots: The Last Wish". But since I can't find any, I decided to just focus
on searching for available storybooks instead since it's quicker for the video
editing in the end. I went to archive.org to search since I can access all
pages of books there. I picked 3 for my friend, alias Virus Ong, to choose as
he agreed on helping me with the narration and voice acting, and in the end,
Aladdin is the winner.
Voice recording
Fig 1.1 All recordings done for the story
I asked my friend to test out the page number he'll reach when he reads the
book in 3 minutes, and we found out that page 1-8 can be read finished within
3 minutes. So, he then started to record himself narrating and voice acting
each of the 8 pages using his iPad. After listening to his first take, I asked
him to do 2 more takes to reduce lip smacking, lowering his voice more for the
Tiger God's line and to record in a more silent environment since background
noise can be heard in one of the takes while he is talking.
Audio editing
|
| Fig 2.1 Noise reduction |
I first started with reducing the noise for the recordings. I choose a section in the audio with only the noise and created a noise print out of it, and then used that noise print to apply the noise reduction effect. The noise reduction percentage is set around 70%-80% depending on the noise level of the recordings. I used the first take of page 1-2 of the story, and second take of page 3-8 as the main recordings. I copied some better parts out from the other recording takes to paste in the main recordings so that I won't have to do the same audio cleaning process in multiple files.
|
| Fig 2.2 Manually adjusting the spikes to be about the same level |
|
| Fig 2.3 Applying dynamic and hard limiter |
After reducing the noise in the recordings, I then manually adjusted some audio spikes that were too loud in the recording to be about the same volume level as the other spikes. Then, I used the dynamics effect the compress the spikes so the volume level at every word is almost constant. After that, I reduced the noise and clicks at every pause in the recordings one by one and used the Autogate feature in the dynamics effect to eliminate any clicks that I might've missed or can't eliminate manually. Hard limiter is then used to limit the volume level of the recordings at maximum -8dB. Input boost is added by a bit to amplify the voice and give it more bass.
|
| Fig 2.4 Refining the voice level |
|
| Fig 2.5 Equalizer graph settings |
Sound layering
|
| Fig 3.1 Multitrack window - Part 1 |
|
| Fig 3.2 Multitrack window - Part 2 |
|
| Fig 3.3 Multitrack window - Part 3 |
After cleaning up the recording, I then proceeded with layering the
recording with suitable music and sound effects. I sliced some parts from
the voice recording to their own tracks so that I can apply different
effects to them. Track automation is also used to control the volume of the
music at different parts of the recordings. The source links to all the
music and sound effects can be found at the bottom of the blog post.
|
| Fig 3.4 Effects rack of Tiger god voice |
|
| Fig 3.5 Pitch shift settings for Aladdin voice |
|
| Fig 3.6 Pitch shift settings for Jasmine settings |
Tiger God's, Aladdin's and Jasmine's voices are pitch shifted a bit so that
they sound more like their character. Full reverb is added to Tiger God's
voice since it is coming from a cave and to make it sound more booming, and
hard limited is added because the audio clipped due to the loudness of the
reverb. The amplify effect was there to test out whether it would make the
voice clearer by amplifying it, but then the effect was turned off since it
isn't needed when the volume of the track can be controlled.
|
| Fig 3.7 Equalizer settings for all the voice tracks |
|
| Fig 3.8 Mastering and Hard limiter settings in master track |
All the voice tracks were applied an equalizer setting to make the sound
more clearer and less muddled, since there was too many bass and too less
treble in Fig 2.5. I also added a DeNoise effect set at 8%-9% to further
remove some noise when the recording starts talking. After finishing with
layering and adding effects to the sounds, a Mastering effect is added to
the master track to make the full production sound clearer. Hard limiter was
also applied to make sure the final volume does not go over -8dB, input
boost was added a little to compensate the loss of volume from limiting
it.
Video editing
|
| Fig 3.9 Premiere Pro workspace |
Fig 3.10 First draft of audio storytelling
After finishing with the audio editing and layering, I exported the audio
and brought it into Adobe Premiere Pro to edit it with the screenshotted
pages of the storybook e-book. Since the pages of the storybook source
aren't very clear, I brightened the screenshots and sharpened them to make
them more vibrant and clearer. Then, I used a transition called "Fold" for
all the pages and "Unfold" for the last page that transition to the "to be
continued" text. These transitions are from the Universe plugin by Red
Giant, I had to use the plugin transitions because for some reason, Premiere
Pro's built in transitions doesn't show during video playback.
Further editing of sound layering
Fig 4.1 Full multitrack window and effects rack of each track
|
|
Fig 4.2 Edited equalizer settings for the voice tracks |
Final Audio Storytelling
Fig 5.1 Final audio storytelling
Feedbacks
Week 11
Overall not bad. Sound effects are ok, though they sound a bit too busy,
maybe choose which of the sound effects to make it softer. The footsteps and
the market sounds can be lower. Make the sound effects and music really
subtle after playing them for a while. Lower the bass of the vocals a bit,
especially the narration to make it consistent with the voice acting. Can do
panning for the sound effects to make more room for the vocals.
Reflection
This project was enjoyable for me to do as I get to try out making a phone
recorded vocal sound somewhat like it's recorded from a mic, and also get to
do vocal and sound effects pairing which felt like editing for a film
production a bit (minus the video). The most time consuming part is of course
the clean up process for the voice recordings as I needed to manually clean it
up at some parts and try a lot of different settings to see which works best.
The sound layering was a bit faster since I got familiar with the process from
Project 1, just that finding the right music took some time because I wanted
solely Arabic inspired music to match the story. But all in all, I got to
learn a lot about audio processing in this project and thank you for my
friend, Virus Ong, for helping me with the narration and voice acting
because I surely will butcher it if I did it on my own.
Sound source
- Egypt Music by TheoJT
https://freesound.org/people/TheoJT/sounds/511198/ - Eastern Rythm Loop _ Kanjira Djembe.wav by ClementPanchout
https://freesound.org/people/ClementPanchout/sounds/572900/ - Eastern Guitar Nylon String 01 - 130bpm - D - New Nation.wav by nnaudio
https://freesound.org/people/nnaudio/sounds/591098/ - Ramadan Time by SergeQuadrado
https://freesound.org/people/SergeQuadrado/sounds/654971/ - arab-group-1.wav by xserra
https://freesound.org/people/xserra/sounds/186703/ - Drum Hit 3.wav by robertmcdonald
https://freesound.org/people/robertmcdonald/sounds/139507/ - Arabic rythm by molleik
https://freesound.org/people/molleik/sounds/250859/ - Arabic 2-4.wav by nemaavla
https://freesound.org/people/nemaavla/sounds/510745/ - arab-drums-6.wav by xserra
https://freesound.org/people/xserra/sounds/186728/ - Landmass / Earth Rumble by el-bee
https://freesound.org/people/el-bee/sounds/363122/ - SAND POUR.wav by nicoproson
https://freesound.org/people/nicoproson/sounds/627070/ - SRS_Foley_Horse_Galloping.wav by StephenSaldanha
https://freesound.org/people/StephenSaldanha/sounds/165532/ - Door_Slam01.flac by corpocracy
https://freesound.org/people/corpocracy/sounds/403355/ - Dropping Plastic Brick in Grit.wav by F.M.Audio
https://freesound.org/people/F.M.Audio/sounds/547584/ - wood_toy_wheel.aif by joseph.larralde
https://freesound.org/people/joseph.larralde/sounds/352423/ - Grab 1 by ihitokage
https://freesound.org/people/ihitokage/sounds/395328/ - Grab 3 by ihitokage
https://freesound.org/people/ihitokage/sounds/395332/ - Opening and closing a big old door.wav by lazymonk
https://freesound.org/people/lazymonk/sounds/214307/ - Crowds.wav by Metzik
https://freesound.org/people/Metzik/sounds/244623/ - Lebanese Arab Market busy people traffic music .wav by Ramston
https://freesound.org/people/Ramston/sounds/262231/ - cicade at nighttime animals 02 by Eelke
https://freesound.org/people/Eelke/sounds/462620/ - Slight cricket singing with silent breeze recorded at night in the desert at Bagdad California by felix.blume
https://freesound.org/people/felix.blume/sounds/651241/ - Bird Whistling, A.wav by InspectorJ
https://freesound.org/people/InspectorJ/sounds/339326/ - Desert silence at night by Diegolar
https://freesound.org/people/Diegolar/sounds/411774/ - Footsteps Dirt 02 by aglinder
https://freesound.org/people/aglinder/sounds/264474/ - Running on beach sand.wav by Noted451
https://freesound.org/people/Noted451/sounds/531014/
Comments
Post a Comment