Swara Ibu

Inspiration

Behind the smile of a new mother often lies a quiet struggle. Many experience exhaustion, emotional instability, and isolation, yet their pain remains unseen. In Indonesia, postpartum depression affects more than 18.37% of mothers, a serious issue that is rarely addressed openly. This alarming reality inspired us to create a solution that goes beyond information and becomes truly present in moments of emotional need. We envisioned technology that does not just respond, but listens. From that vision, Swara Ibu was born as an emotional companion that understands mothers through their voice and helps them feel seen, heard, and supported.

What it does

Swara Ibu is a mobile application that uses artificial intelligence to analyze the voice of postpartum mothers and detect their emotional state in real time. Each day, the mother is invited to perform a voice check-in by answering 5 reflection-based questions. Her voice is recorded and analyzed to determine emotional indicators such as sadness, anxiety, fatigue, or stress. The app provides an emotional summary, a stress level score, and personalized suggestions powered by a large language model called Gemini.

When the system detects critical signs or harmful speech content, an emergency alert is immediately sent to the user's chosen companion. This companion, often a spouse or trusted family member, is connected through a unique code and can access a dedicated view of the mother's emotional condition and history. This allows them to provide timely and meaningful support.

How we built it

The frontend of Swara Ibu was built using Jetpack Compose to ensure a seamless mobile experience. We used FastAPI for the backend to handle API requests and data flow. Voice recordings are processed using OpenAI Whisper for transcription, while Librosa is used to extract audio features such as pitch, tone, and tempo. These features are analyzed by a custom machine learning model we trained using over 100 emotion-labeled audio samples from Indonesian-speaking mothers.

For logic and orchestration, we used Firebase Cloud Functions to connect the frontend, backend, and analysis engine. Notifications and alerts are sent through Firebase Cloud Messaging. Gemini serves as the suggestion engine that tailors advice to the mother's emotional state. We designed the app to support 2 user roles: mothers and companions, both of whom experience the app through separate but synchronized workflows.

Challenges we ran into

Training the emotion recognition model proved to be a major challenge. Despite collecting over 100 audio recordings for training, the model's accuracy reached only about 70%. This limitation highlights the complexity of interpreting emotions from voice, especially when factoring in cultural context and language-specific nuances.

We also encountered technical obstacles in implementing real-time emergency alerts. Creating a reliable notification system that could deliver instant and accurate alerts to a companion's device required significant testing and adjustments. Ensuring the system was fast without being overly sensitive was another balancing act we had to solve.

Accomplishments that we're proud of

We are proud to have built a working prototype that combines emotional AI, speech recognition, and real-time alerts into a functional mobile application. One of our greatest accomplishments is successfully enabling 2 types of users, mothers and companions, to interact in a way that supports emotional care and safety. We also achieved smooth integration between various components, from audio feature extraction to language model-generated suggestions.

Most importantly, we are proud of designing an experience that is centered on empathy. Swara Ibu is not only a technical product but also a human-focused solution for one of the most emotionally vulnerable periods in a woman’s life.

What we learned

Throughout the development process, we learned that building with empathy is just as important as building with precision. Addressing the emotional needs of postpartum mothers taught us the value of careful design, cultural sensitivity, and continuous feedback from users. We also learned the importance of cross-disciplinary collaboration, where technical, design, and behavioral perspectives must align to create a meaningful experience.

What's next for Swara Ibu

In the next stages, we plan to expand and improve the accuracy of our emotion recognition model by collecting more diverse voice data from Indonesian mothers. We aim to connect the app with licensed mental health professionals, enabling follow-up care when signs of emotional risk appear. In addition, we will develop features such as daily reflections, positive affirmations, and long-term emotional tracking. Our goal is for Swara Ibu to be more than just an app. It should become a daily companion that listens and responds with compassion.