This is a submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass
What I Built
EchoQuest is an AI-powered nature sound discovery app designed to encourage people to step outside, listen more carefully, and reconnect with the natural world.
We spend so much time looking at screens that we sometimes forget to pay attention to what's happening around us.I wanted to explore how AI could make outdoor exploration more engaging instead of giving people another reason to stay online.
The idea is simple: go outside, listen to the world, and turn what you hear into a discovery.
With EchoQuest, users can record sounds around them and analyze them using an AI audio classification model. The app displays the model's predictions alongside nature-focused insights, turning an ordinary walk into a small exploration mission.
To make the experience more engaging, I added a few lightweight gamification features:
π± Sound Discovery: Record an outdoor sound and explore the model's predictions.
β¨ XP Points: Earn points for supported nature discoveries.
π Nature Journal: Review previous discoveries and track your progress.
π Achievement Badges: Unlock milestones as you continue exploring.
EchoQuest is built for curious people, students, nature lovers, and anyone who could use a little motivation to take a break from their screen.
The goal is simple: step outside, listen closer, and explore. πΏ
See EchoQuest in Action
1.Sound Prediction
Start by recording a sound from your surroundings. EchoQuest processes the recording and displays the model's predictions.
Once the analysis is complete, users can explore the results and see whether the sound offers an interesting nature discovery.
2. Nature Journal and Badges
Discoveries can become part of an ongoing journey. The journal keeps track of previous results, while XP points and achievement badges add a sense of progress.
Visit the Demo & Try It Yourself
Demo
π Live demo:
π₯ Video demo:
Code
π» GitHub repository:
Gulnur-Altan
/
echoquest
πΏ An AI-powered nature sound discovery app that turns outdoor walks into quests with audio classification
πΏ EchoQuest β Touch Grass, One Sound at a Time
Turn your next walk into an AI-powered nature discovery quest.
ποΈ Audio Classification Β· π± Nature Discovery Β· β¨ Gamification
Built for Hacktoberfest Open-Source AI Challenge β Week 1: Touch Grass
π About
EchoQuest is an AI-powered web application that encourages people to step away from their screens and reconnect with nature through sound.
Record the sounds around you, explore AI-generated audio predictions, collect discovery points, unlock achievement badges, and build your own nature journal.
The idea is simple: use AI to inspire people to experience the real world, not spend more time staring at a screen.
β¨ Features
- ποΈ Record Nature Sounds β Capture environmental audio directly in your browser.
- π€ AI Audio Classification β Analyze recordings with a pretrained audio classification model.
- π Nature Insights β Explore nature-focused information based on recognized sound categories.
- β Experience Points (XP)β¦
The project source code is available on GitHub. Contributions, suggestions, and bug reports are welcome!
How I Built It
EchoQuest is built with Next.js, React, TypeScript, and Transformers.js.
For audio classification, I use the pretrained Xenova/ast-finetuned-audioset-10-10-0.4593 model. The application records audio through the browser, preprocesses the recording for inference, and displays the model's top predictions. Rather than sending recordings to a paid, hosted inference API, EchoQuest runs the classification pipeline in the user's browser.
The main technical components include:
- Audio recording: The browser's MediaRecorder API captures audio.
- Audio preprocessing: The recording is decoded and resampled to the model's expected 16 kHz sample rate.
- AI inference: Transformers.js runs the pretrained audio classification model.
- Nature-focused interpretation: The app uses model predictions to provide relevant exploratory insights.
- Progress tracking: Browser local storage persists the nature journal.
- Gamification: XP points and achievement badges encourage continued exploration.
One important design decision was to keep the core inference process in the browser rather than depend on a paid, hosted AI inference API. This reduces reliance on external inference services and keeps the audio processing on the client side.
Note: The model is a general audio-event classifier, not a dedicated wildlife species identification system. Its predictions are exploratory hints, not definitive scientific identifications.
Why Does Open Innovation Matter?
Open innovation was important to this project because it made experimentation with AI more accessible.
By using a pretrained model and the JavaScript machine learning ecosystem, I could focus on integrating audio classification into a web application and designing an experience around it, without having to train a model from scratch or depend on a proprietary inference API.
I also wanted to explore how open tools can help developers turn an idea into something people can actually use.
EchoQuest could grow through contributions from other developers. Possible areas for improvement include sound classification evaluation, accessibility, testing, nature-focused insights, and the overall discovery experience.
For me, the interesting part is not just getting an AI model to produce a prediction. It's exploring how that prediction can become part of a useful, engaging experience.
For a project about reconnecting people with nature, I like the idea that the technology behind it can also be explored, inspected, and improved by a community.
What I Learned and What's Next
For me, the most challenging part of EchoQuest wasn't integrating the AI model or writing the code. It was turning the model's predictions into a meaningful game experience. Designing the XP system, badges, and Nature Journal meant finding ways to reward real nature discoveries without treating every AI prediction as a valid discovery.
Ironically, the biggest challenge was actually getting outside and listening to different sounds! π
I learned that building an AI-powered app isn't just about making a model work; it's also about creating an engaging experience that encourages people to spend more time outdoors and notice the world around them.
Next, I'd like to make discoveries more meaningful and improve the reward system.. πΏ
Top comments (4)
Really nice work, GΓΌlnur! πΏ
I love the idea of using AI to encourage people to spend less time staring at screens. Running audio classification directly in the browser is also a thoughtful architectural choice.
I took a look through your GitHub repository, and one small detail caught my attention in getBestNatureMatch(). It selects the first matching nature category from the predictions without checking the confidence score. This means even a low-scoring nature prediction could earn XP.
Since you mentioned that rewarding meaningful discoveries was one of the trickiest parts, a confidence threshold (ideally calibrated against real recordings) might be an interesting next step.
I also noticed that analyzing the same recording again can award additional XP, so some form of duplicate-discovery handling could make the progression system more meaningful.
Just a couple of observations, not criticisms. Youβve built a lovely foundation, and I really like the direction youβre taking it.
Keep building, and keep exploring! π±π
Thank you so much for taking the time to look through the project and share such thoughtful feedback βΊοΈ Youβre right about both points. Confidence thresholds and duplicate-discovery handling are definitely things Iβd like to improve to make the XP system more meaningful. I really appreciate you pointing these out ,it gives me a clearer idea of what to work on next
Layout wow!
thanksβΊοΈ
Some comments may only be visible to logged-in visitors. Sign in to view all comments.