Inspiration
Ever been stuck in a dry conversation full of painful, awkward pauses? Fear no more! mood FM provides your own personal mixtape (a personalized 'radio channel,' if you will), to play background music suited to the nuanced mood of your conversation. mood FM gives you the perfect movie soundtrack for each moment of your life to give you the main character energy you deserve.
What it does
When you choose to “read the room,” mood FM listens to your conversation and interprets the contents of your speech to determine the mood you’re in. Then, it finds a song to complement that mood automatically so you always have the perfect soundtrack playing in the background. mood FM also tracks your emotions on a daily and weekly basis to help you spot potential trends and ensure you’re taking care of yourself.
How we built it
1. Wire-framming and Front-end As a medium of use, it was necessary to build a front-end component to our application as a way to prompt the user to record input, and to relay a specialized playlist based on their mood. Both preliminary and high-fidelity wireframes were designed and validated in Figma before using React.js to bring our ideas to light.
2. Voice to Speech
In order to “read the room,” our project needed to be able to recognize voices and detect speech. We achieved this conversion from voice to speech by employing AWS Amazon Transcribe API. Real-time audio is captured using the sound device library by the device microphone, chunked into audio snippets, and streamed to the cloud for transcription.
3. Speech to Mood
For our main character energy flair, wielding NLP, mood FM understands your mood from your speech using Cohere. Exploring the classification Cohere playground and the Product Review Sentiment, we created training data for 5 mood labels: happy, sad, chill, angry, and stressed. Examples of training data phrases include “I love Hack the North,” (happy) and “Coding is scary” (stressed).
4. Mood to Music
To go from mood to groove, we used a variety of get, post, and put Spotify APIs to play music that best fits your setting and wellness. From blasting Super Bass when you feel on top of the world to chilling with the Up theme song, mood FM playback knows what’s on your mind – it’s like telepathic personalization!
5. Music to Analytics
Along with a personalized playlist, mood FM helps you track your sentiments over time with statistics including daily mood breakdown and 7-day mood trends. With this data handy, users can make more conscious observations of their mood trends. Whether you track your stress levels during finals season or to discover new 3 am mellow tunes, make your decision using data-driven analytics.
Challenges we ran into
One challenge with ideating a multi-step product is finding the right tools for each task. Through research, and asking the amazing Hack the North mentor, our team pinpointed what we thought would be best for the job. Consequently, getting three separate APIs to collaborate and achieve a unifying goal introduced a slew of challenges. For instance, dataframe manipulation was required to ensure the output from one API was compatible with the other and several helper methods were developed to interpret and translate the different results along our mood-centric music pipeline.
Another challenge we ran into was ensuring effective communication when determining how and front-end and back-end components would connect together to support a seamless user experience. Given the time constraint of the hackathon, both the frontend and backend components of the projects were being developed in parallel and thus, one of the learning curves was how the final puzzle pieces of our project fit together given that they were completed independently.
Accomplishments that we're proud of
Learning 3 different API this weekend (starting ground 0 of API calling knowledge), while brining hours troubleshooting, was extremely rewarding and provided us the knowledge to implement these skills in future projects (we even used a Cohere text API - putting us in the run for a Cohere API hackathon prize!)
High-fidelity app wireframe designed in Figma
Our team’s first dive into using React for app development
What we learned
During the hackathon, we learned a variety of technical, design, and entrepreneurial skills. New knowledge gained from our brand end to end app developing, API calling, We also learned more about the interrelated nature between technology and wellbeing. Our team was constantly thinking of questions such as what differences exist between what music people need/want and how often tracks should change. These considerations helped us design features that make the user feel good using them!
What's next for mood FM
Our next steps include asking mood FM to play certain categories and adding more emotions for more nuanced song picks. To further emphasize the necessity of wellness considerations in our everyday interactions with technology, mood FM plans to integrate music-based breathing exercises and mindfulness-promoting media resources. For example, considering how breathing can also impact mood, a “breathe beats” feature can be implemented to manipulate the user’s breathing rhythms (ie. slower or deeper) for a healthier lifestyle.
Log in or sign up for Devpost to join the conversation.