17.06.2025 - 01.07.2025 (Week 9 - Week 11 )
Janaan Ahmed (0353333)
Minor Project/ Bachelor of Design (Hons) in Creative Media
Project 3: Audio Storytelling
Lectures
Week 9: Noise Reduction & Clean Up
I missed class but from my peers I learned that this week was about learning to reduce noise in our voice recordings, through 2 methods:
Noise Reduction
- We select part of the clip with the noise and press [Shift] + P or Capture Noise Print.
- Then we select the entire clip
- And adjust the settings
 |
| Fig 1.1: Noise Reduction |
Compression (Manual and Automation)
We can manually adjust the volume of different sections in a clip but we can also go to Amplitude and Compression > Dynamics.
 |
| Fig 1.2: Manually adjusting volume peak |
Dynamics Tab
AutoGate
- A tool that automatically mutes or lowers quiet parts of your audio. It only lets sound through when it's loud enough.
- If the sound is loud enough, the gate opens --> sound passes through.
- If the sound is too quiet, the gate closes --> sound is blocked or reduced.
- It's used to:
- Cut out background noise between words.
- Remove breaths or silence or pops in voice recordings.
 |
| Fig 1.3: Dynamics FX rack |
Settings:
- Threshold: How loud the sound must be to open the gate.
- Attack: How quickly the gate opens.
- Release: How quickly the gate closes after the sound gets quiet.
- Hold: How long the gate stays open before closing.
Compressor
- Automatically turns down loud sounds so your audio is more even.
- If a sound is too loud, it gets reduced
- Quiet parts can be boosted with makeup gain
- Use to balance voice and smooth out volume jumps.
- Make audio sound cleaner and fuller
- Threshold: When to start compressing.
- Ratio: How much to reduce loud sounds.
- Attack: How fast it reacts.
- Release: How fast it stops.
- Makeup Gain: Boosts overall volume.
Note: Before using compressor make sure to manually adjust spikes in your clip! Otherwise the compressor will reference this highest peak which will thus affect other peaks.
Instructions
Project 3: Audio Storytelling
For this project we have to choose a popular fairy-tale story of our
choice and create and audiobook of about 2-3 minutes.
PROCESS:
1. Script + Dialogue breakdown
Initially I selected Goldilocks & the Three Bears thinking the
narration might be fun, but I found the dialogue a bit repetitive. So
I decided to do The Little Mermaid from Disney since it has
more dynamic voice acting. The story follows up to the point where
Ariel gets legs and swims to the surface. I watched some video clips
from the movie and tried to match the dialogue.
2. Voice Recording
There are 4 distinct voices for this audiobook which I recorded myself using a phone:
- Narrator
- Ariel
- King Triton
- Ursula
I record the dialogue and narration for all 4 parts accordingly,
referencing some clips from Youtube to kinda mimic the characters'
enunciation and speaking manners. I gotta say I had way too much fun
with this.
I also made an additional recording of the entire script in one way to
get the timing and pace right.
Clips referenced:
Voice recordings:
3. Editing Audio
I use my reference clip to identify the parts where different
characters jump into the narration. Then I string the individual
voice recordings together. I edit each character's voice
using pitch shifter and chorus.
Narrator
- Captured noise print and reduced noise to minimise white noise.
- Used Dynamics > Autogate and set the threshold level to match the frequency of unwanted sounds such as breathing.
- Used Dynamics > Compressor to reduce high streaks (uneven volume)
 |
| Fig 2.1: Windows showing denoise and compressor |
I also notice that there was a lot of sibillance or 's' sounds in my audio so looked up a tutorial to use the DeEsser.
 |
| Fig 2.2: Identifying problem frequency with Frequency Window |
Using Shift + D we get to view the frequency window for the selected clip. After spotting the problem frequency. From the side panel we see that it's roughly between 7k to 10kHz. Next we go to the DeEsser FX panel and adjust thee centre frequency, ensuring we also keep the threshold gain somewhat in the middle so it only cuts the s part.
 |
| Fig 2.3: Change Centre Frequency to the inbetween frequency of the range |
Note: Repeated a similar process for all the remaining voices (noise reduction and compressing).
In the multitrack session, I then applied the following FX to the Narrator:
- Pitch Shifted down 2 semitones to alter my voice into a more mature story-teller kind of voice you hear in infomercials.
- EQ: Boosted band 1 and 5 for increased bass and clarity and reduced mid-range to reduce muddiness.
- Hard Limiter: To avoid potential clipping
Ariel
FX applied:
- EQ: Decreased midrange and increased high bands for better clarity and sharpness
- No pitch change applied.
King Triton
FX applied:
- Pitch shifted down 8 semitones to make a convincing-enough manly voice
- EQ: Boosted high bands for increased clarity
- Reverb: King Triton's dialogue is in a cavern so we needed to increased the spread
Ursula
FX applied:
- Pitch shifted down 4 semitones for alto voice
- EQ: Boosted band 1 for increased bass, and high bands for clarity
- Hard Limiter: To avoid clipping
4. Compiling Visuals (Premier Pro)
I find visuals that match my script and downloaded relevant
screencaps from the movie itself: https://animationscreencaps.com/the-little-mermaid-1989/page/1
 |
| Fig 2.4: Library with imported movie screencaps on Premier Pro |
After that I import the dialogue/narration audio file from Audition to
time the scenes properly and add video transitions and slight effects
(such as VR Light Leaks transition) for impactful scenes. This was particularly where King Triton destroys Ariel's tavern.
 |
| Fig 2.5: Applying transitions |
But overall I try to keep the transitions limited, minimal and cohesive so it doesn't become distracting for the viewer (cross dissolve). The audio is the selling point here after all.
5. Adding SFX
I import the newly compiled video into Audition to figure out where to
put the sound effects for each scene. Since it is an audiobook, the
SFX don't need to be at the forefront but rather a supporting element
to the story. This is basically like the environment soundscapes we
did earlier. I pick and choose 1-2 defining elements from each visual
to animate through sound and delegate them to different bus tracks so it matches that given environment.
 |
| Fig 2.6: Multitrack window |
All sound effects are downloaded from
freesounds. I build on the scenes
from the base up meaning ambient sounds, and then foreground sounds.
Additionally we get to add music so with much difficulty finding ones
that match, I incorporate those to add to the emotional
weight/language of the scene.
 |
| Fig 2.7: Mixer |
As a finishing touch, added a Hard Limiter to the Mixer to prevent peaking and audio clipping. We then export the mixdown to Premier Pro and put audio and video together.
FINAL OUTCOME
Feedback
-
Reflection
I think my days playing dolls with my baby cousin sisters came through for this project in terms of voice acting haha! This was a fun project through and through where we get to combine what we learned from auditory imaging as well. Though maybe a little extensive since there are lots of parts to it, from voice recording, compiling the visuals, and adding the SFX. But honestly it's a fun way to learn more about technical aspects in Audition such as the noise reduction, de-esser etc. I'd be lying if I said I didn't get a little overwhelmed though because honestly there's a billion different ways to interpret an image/story and apply sound that one finds fitting. And it was really hard to find royalty free music that matched the fairy tale + vintage Disney score. But I'm quite happy with the end result though with later listens I think I could've reduced the SFX as I observed it overpowered the narration at times. But we live and we learn.
Comments
Post a Comment