Week 9 - Week 11 : 30 May 2023 - 18 Jun 2023
Tan Yi Yun 0345559Bachelor of Design (Hons) in Creative Media
Project 2/Audio Storytelling
Students are to pick one popular fairy-tale story of their choice and
create and audiobook. The task is a step up from auditory imaging whereby
now the students not only need to think about suitable soundscape, but
they need to record the voice with the suitable voice acting and bring it
all together with sound effects to highlight and enhance the storytelling
experience. Imagine that this is a show for radio.
Here are some techniques that can be applied to edit our raw voice
clips.
Noise reduction
Select part of the clip
Effects/Noise reduction/Capture Noise Print
Select the entire clip
Effects/Noise reduction/Noise reduction (process)
|
|
Figure 1.1, Noise reduction (process), Week 9 (30/5/2023) |
Before doing the compression (automation), we should do it manually first
to reduce a few highest "spikes".
|
| Figure 1.2, Compression, Week 9 (30/5/2023) |
Adjust the threshold to match the frequency of the unwanted sound (breathing).
|
| Figure 1.3, AutoGate, Week 9 (30/5/2023) |
Storyline
The story I have chosen is "Rumpelstiltskin", a German fairy tale. The
story is about a little imp who spins straw into gold in exchange for a
girl's firstborn child.
Storyline, Week 9 (2/6/2023)
Images
I took screengrab from this read aloud video: https://youtu.be/Eh1vjrr2HtU
|
||||
|
Figure 1.4, Images, Week 9 (29/5/2023) When I was doing the compilation, I realized that the images were too blurry. In the end, I looked for another video and exported it in PNG sequence to preserve the quality (although I do not like the art style haha). This is the video source: https://youtu.be/sMEC4xOSQG0
|
I recorded my voice using Voice Memos. I narrate it with different emotions
when different characters speak. I also reduce the high spikes to make the audio sounds more
consistent.
|
|
Figure 1.6, Voice editing, Week 10 (9/6/2023) |
Voice-over (draft), Week 10 (9/6/2023)
I used pitch shifter and parametric equalizer to change the pitch of the
miller's voice.
|
| Figure 1.7, Parametric equalizer, Week 10 (9/6/2023) |
|
| Figure 1.8, Pitch shifter, Week 10 (9/6/2023) |
When I was trying to edit the voice of the king, I found the tone of voice
and the pitch is too similar (because I record my voice for both
characters). Instead of finding a friend to help me record, I utilized the
AI voice and searched for a suitable voice to match the characters'
personality as I wish the audio to sound as convincing as
possible.
One of the issues I encountered is I was forced to change the original
script because of the word "die" presents in my script. Next, I also spent a
few time to scan through the suitable narrator: they have different genders,
age, style, talking speed and et cetera. It is an experimental process
because I have never tried AI voice before. There is also a problem of AI
not following the script exactly what I have written. This might due to the
type of narrator I have chosen, since the problem resolved when I changed to
another narrator.
|
| Figure 1.9, Generating AI voice, Week 10 (9/6/2023) |
|
| Figure 2.0, Generating AI voice, Week 10 (9/6/2023) |
Audio Compilation (AI Voice)
Audio compilation (AI Voice), Week 10 (9/6/2023)
Then, it is time to add sound effects. I downloaded the sound effects from
Freesound. Some of the background music is taken from YouTube that are free
copyrighted.
|
| Figure 2.1, Sound effects, Week 10 (9/6/2023) |
Draft 1
The major problem of this draft is the quality of the images, and the usage
of AI voice (which doesn't fulfill the requirement). After the feedback
session, I made a few changes, not just the images and voice, but also the
sound effects to match the new visuals.
Draft 1, Week 11 (16/6/2023)
During the consultation, Mr. Razif gived a suggestion to tune down the "s"
sound (which is also called sibilance). There are tools called
De-Essers that help with this problem.
|
|
|
|
|
|
I asked my friends to record the dialogue of these two characters. Here is the original clip:
The little man, Week 11 (16/6/2023)
Stretch and pitch was applied to increase the talking speed and tuned up
the pitch to sounds like an imp.
|
| Figure 2.4, Stretch and pitch, Week 11 (17/6/2023) |
The clip below is an original clip of the king's dialogue. I applied pitch
shifter to tuned down the pitch so that the voice sound thicker.
The king, Week 11 (16/6/2023)
|
|
|
Next, I also slightly tuned down the high pitch and the bass to the entire
narration clip. I created a bus and named it "narration" because I wanted to
apply certain effects purely on the narration and dialogue.
|
| Figure 2.6, Bus track, Week 11 (17/6/2023) |
I applied DeEsser Hard, DeSher and adjusted Parametric Equalizer so that
the voice-over will sound more pleasing to the ears.
|
| Figure 2.7, DeEsser Hard, Week 11 (17/6/2023) |
|
| Figure 2.8, DeSher, Week 11 (17/6/2023) |
|
|
Figure 2.9, Parametric Equalizer, Week 11 (17/6/2023) |
|
| Figure 3.0, Sound effects, Week 11 (17/6/2023) |
|
|
|
Video compilation
Last but not least, the compilation of the visuals and the audio was done
in Adobe Premiere Pro. I also added some simple animation like zooming the
images.
|
|
Figure 3.2, Sound effects, Week 11 (17/6/2023) |
Final Submission
Final submission of auditory storytelling, Week 11 (17/6/2023)
FEEDBACK
Week 11 :
The narration is good. Use human voice to replace the AI voice. This is to
learn the fundamental skills of voice editing, which is part of the project
requirement. But of course it's not wrong because of the advancement of
technology, which you can use it for future projects. Also, enlarge the voice to reach between -6 and -3. The current "s" pronunciation
is very sharp, you can apply De-Esser to reduce the spikes.
REFLECTION
I have learned a lot about the art of storytelling and the technical
aspects of audio production. It actually brought back some memories of the
storytelling competitions in primary school, in which I was given this story
in Standard 5 and was trained to participate in a storytelling competition.
This is the reason why I chose to retell this story. Back in the days I
didn't have much access to the Internet and I was curious to see how
"Rumpelstiltskin" looks like. Now this project allowed me to search for the
visuals and I found it very interesting to look at different versions of the
character. Besides, this project also allows me to explore the tool of AI
Voice. I was aware of the advancement of AI and it was fun to try and
experiment although in the end it should not be used for this project. By
exploring noise reduction methods, I learned how to minimize background
noise and distractions, like turning off fans or air conditioners, and
choosing a quiet room. Additionally, I experimented with positioning the
microphone correctly to capture clear and high-quality audio. For the sound
effects part, I did not encounter much issues since most of them are
available on Freesound and YouTube. I spent most of the time to tune the
voice-overs to make them sound believable. Some of the raw clips cannot be
used due to the problem of recording devices. The visuals part supposed to
be easy, but then I wasted a lot of time because I couldn't find a set of
visuals that match the story. The visuals I have chose initially were blurry,
so I had to look for other sources and download the new video. I rendered it
to PNG sequence, resulting in hundreds of images, and filtered out the scene
that did not match the storyline. In the end, I am happy to see the final
results.