For this Hackathon, I wanted to explore audio based Sentiment Analysis using DeepGram. I wanted to use DeepGram to transcribe pieces of audio, and then use a Sentiment Analysis library to analyze the transcribed texts in terms of their sentiment.
I did a bit of thinking about what pieces of audio I'm going to perform Sentiment Analysis on and decided that it would be a good idea to make a Python app that performs Sentiment Analysis on Marques Brownlee's
smartphone reviews from the past 3 years to find, each year, what smartphones he liked, what smartphones he didn't like as much, as well as his sentiment towards the different smartphone brands.
Marques reviews a few different brands but I've chosen the ones that have been reviewed the most by Marques so that I have enough data to work with. These brands are: Google, Samsung and Apple. I'll only be referring to these when talking about smartphones and brands in this article.
I know what you might be thinking: "But Sentiment Analysis isn't very accurate and can make misjudgments on someone's sentiment towards something such as a smartphone!". I know that Sentiment Analysis has it's limitations.
Nevertheless, I've used the best Sentiment Analysis Library in my experience to try and programmatically make the best assessment of his reviews.
Submission Category: Analytics Ambassadors
Building the App 🛠
The app uses 3 scripts to perform Sentiment Analysis on Marques' smartphone reviews from each year.
Each script does the following:
- Using the TextBlob library, it performs Sentiment Analysis on Marques Brownlee's review videos from the year represented by the script(2019, 2020, 2021) that I downloaded in
.mp3format using this web tool. - It prints Marques' favorite and least favorite smartphone of the year based on the Sentiment Analysis.
- It prints Marques' average sentiment towards the different smartphone brands throughout the year.
- It displays his favorite brand of the year based on the sentiment analysis.
- It displays a bar graph of Marques' sentiment towards each smartphone he'd reviewed that year.
The reason why I made 3 different scripts analyze smartphone reviews from 3 years separately is because I felt that the bar for a great smartphone raises every year. Therefor it wouldn't be fair to compare, say, an Iphone review from 2019 to an Iphone review from 2021.
You can find the Github repo for this project here: I wanted to explore audio based Sentiment Analysis using DeepGram. I wanted to use DeepGram to transcribe pieces of audio, and then use a Sentiment Analysis library to analyze the transcribed texts in terms of their sentiment. I did a bit of thinking about what pieces of audio I'm going to perform sentiment analysis on and decided that it would be a good idea to make a Python app that performs sentiment analysis on Marques Brownlee's
smartphone reviews from the past 3 years to find, each year, what smartphones he liked, what smartphones he didn't like as much, as well as his sentiment towards the different smartphone brands
Marques Reviews a few different brands but I've chosen the ones that have been reviewed the most by Marques so that I have enough data to work with. These brands are: Google, Samsung and Apple. I'll only be referring to these…
matveynikon
/
MKBHD_Smartphone_Sentiment
This is a project that makes an effort to analyze Marques Brownlee's sentiment towards different smartphones through his youtube reviews.
The Analysis
Here's the analysis that my Python scripts have performed!
Analysis on Reviews from 2019
Here's the graph of Marques Brownlee's sentiment towards the smartphones he'd reviewed in 2019:
Average smartphone sentiment for each brand:
Google: 20.05
Apple: 14.89
Samsung: 18.53
Favorite smartphone: Galaxy Note 9
Least favorite smartphone: Galaxy S10e
Analysis on Reviews from 2020
Here's the graph of Marques Brownlee's sentiment towards the smartphones he'd reviewed in 2020:
Average smartphone sentiment for each brand:
Google: 17.14
Apple: 14.75
Samsung: 16.47
Favorite smartphone: Galaxy Note 10+
Least favorite smartphone: Iphone 11
Analysis on Reviews from 2021
Here's the graph of Marques Brownlee's sentiment towards the smartphones he'd reviewed in 2021:
Average smartphone sentiment for each brand:
Google: 17.26
Apple: 17.84
Samsung: 17.36
Favorite smartphone: Pixel 6/6 Pro
Least favorite smartphone: Galaxy S21 Ultra
Marques' Brand Sentiment Over the Last 3 Years
Here's Marques Brownlee's smartphone brands ranked by average product sentiment over the last 3 years:
- Google(54.45)
- Samsung(52.36)
- Apple(47.48)
Conclusion
I hope that you found this DeepGram project pretty cool!
Byeee👋
Top comments (0)