Voice Dairy

The Problem
As a high school student, I can clearly see the deterioration of mental health amongst my peers, which is concerning. Upon conducting some research, I found that over 1 in 4 adults worldwide have suffered from some sort of mental ailment. That's over 2 billion people. This percentage dramatically increases amongst teenagers, where about 50% of all teens face mental problems like severe depression and anxiety. Mental health is one of the most pressing issues we face in modern society. Unlike physical health, mental health is often downplayed.
The Solution
The solution to alleviate a substantial portion of these problems is surprisingly simple: Journaling. The University of Rochester Medical Center uncovered that journaling aids in controlling symptoms and improving mood in mental health conditions by providing an outlet for relieving stress and worry. A comprehensive study found that writing about emotionally charged experiences for just 15–20 minutes a day can significantly improve both physical and psychological health.
Moreover, teenagers, a particularly vulnerable group for mental health issues, can benefit immensely from a consistent journaling habit. Teenagers who incorporate journaling in their routine have demonstrated marked reductions in symptoms of depression and anxiety. Thus, the practice of journaling is an evidence-backed strategy to address mental health issues experienced by over 2 billion people worldwide.
Why don't more people journal?
Traditionally, journaling involves writing, which comes with its share of challenges:
- Starting can be daunting: A blank page can be imposing, and getting started might be difficult.
- Writing blocks: These can make traditional journaling a tedious process, often leading to demotivation.
- Perception problem: Some people consider journaling mundane and boring.
- Persistence: While it's hard to get started with journaling, it can be even more challenging to maintain the habit continually.
An Innovative Solution: Voice Dairy
To overcome these reported challenges with traditional journaling, Voice Dairy was built to provide a superior, conversational journaling experience. Rather than writing thoughts down, Voice Dairy enables natural conversations that are much easier for almost everyone to engage in. Humans are often better communicators than they are writers. Hence, it is easier for us to talk to a friend about how our day went and reflect in that manner than writing it onto a document or piece of paper. Voice Diary aims to provide that conversational and natural feel to journaling while maintaining the intimacy and honesty of private thought.
Here's what Voice Dairy offers:
- Facilitates talking, which is generally less challenging than writing.
- Emulates a conversation with a friend to help users express their thoughts.
- Ensures that these conversations are neatly logged as journal entries, creating a record for future reflections.
Every element of Voice Dairy has been designed with consideration of the user, right from the interface, to the specific features, and even the back-end technology used to run it. Each element contributes to transforming journaling into an interactive and intuitive experience.
Voice Dairy Interface
Main Page
The Main Page serves as the nerve center of user engagement with the Language Learning Model (LLM). This is where users can record their thoughts in conversation form with the AI, and these interactions are neatly converted into journal entries.

Key features of the Main Page:
- The LLM excels in asking probing questions about the user's day and experiences, guiding them on a reflective journey of their lived moments.
- Past journal entries aren't forgotten but re-visited through follow-up questions, keeping the user engaged and ensuring no topic is left incomplete.
- A dropdown option is available for users to adjust the agent's demeanor, catering to their emotional state.
- Multimedia capability: The user can upload images via the modal on the left, further enriching their journal entries.
- The user enjoys the flexibility of either voice interaction or typing, and they can also listen to the LLM to enhance the natural interaction.
History Page
The History Page provides a neatly organized interface to store and review previous journal entries. Journal entries are saved with date landmarks and can be easily accessed.

Exciting features of the History Page:
- Automatic synopsis generation: Each diary entry comes with a summarized version of the conversation for quick reference.
- Emotional tagging: The software identifies the top three emotions from the conversation and automatically tags them for easy tracking.
- Sentiment score: Every conversation is assessed for its overall positivity and assigned a score, color-coded for visual ease.
- A preview of uploaded images is included in the entry card.
- Diary entries can be expanded for more detail, displaying all uploaded images, a comprehensive summary, conversation transcript, and audio playback capability.
Metrics Page
The Metrics Page focuses on utility and motivation, providing insightful data about user entries. Progress charts depict users' emotional evolution over time, aiding in tracking improvements or identifying prolonged slumps.

Intriguing features of the Metrics Page:
- Histograms display the frequency of tagged emotions.
- The page provides statistics such as the number of words written, the average day score, total entries, streak comparisons with other app users, and calendar streaks.
The Backbone of Voice Dairy: Technologies Employed
Voice Dairy is powered by a combination of state-of-the-art technologies:

- The application UI was engineered using ReactJs, DaisyUI, and Tailwind
- The LLM was designed using the Google Vertex Agent Builder, which allows for the creation of three different agents to facilitate diverse conversation styles. Each mood (optimistic, therapeutic, and inquisitive) that can be selected in the chat dropdown is a different agent with different instructions.
- The Data Store for the Vertex AI agent stores all previous conversations the user has had with the LLM. This feature allows the LLM to revisit past experiences, making the journaling process more effective.
- ElevenLabs API is utilized to give the model lifelike voices.
- Conversation summaries and tags are efficiently generated using the OpenAI API.
- Charts and graphical representations are created with Chart.js.
Log in or sign up for Devpost to join the conversation.