生成式AI應用
Symphony AI
Poor sleep and insomnia affect every age group, yet most tools ignore real-time changes in emotion and physiology. People need a drug-free solution that adapts the moment their bodies do.
Our journey began at NVIDIA GTC, where we identified the biggest gap: harmonizing physiological and emotional signals in real time to induce natural sleep.
Each night, users engage in a “resonance ritual.” Symphony AI reads heart-rate variability, breathing rhythm, micro-movements, and inferred mood, then composes music that evolves as the body relaxes.
A multimodal agent ensemble—Signal Analyzer, Emotion Inference, Generative Composer, and Evaluator—runs on an edge gateway. Latency stays low, and all raw data remain on device.
Sensor data flow through secure pipelines into our agent layer, which drives on-device music generation. Cloud storage contains only anonymized model updates, maintaining HIPAA-grade privacy.
The Evaluator scores each track against overnight metrics, labels what worked, and triggers incremental fine-tuning so tomorrow’s playlist is even more effective.
Key, tempo, and harmonic texture shift continually, guiding users from wakefulness into deep, restorative sleep without manual intervention.
We operate a B2C subscription with optional B2B licensing for hospitals, sleep clinics, and wellness resorts.
Generative AI Projects
Our team focuses on researching the applications of generative AI in the medical and health sectors. We emphasize the importance of preprocessing, refinement, and summarization to create optimal products. Through prompt engineering optimization, we enhance the utility of retrieved data, ensuring better outcomes even with less-than-ideal inputs. We specialize in clinical data annotation and preprocessing, improving human-centric model platforms. Additionally, generative AI aids in generating synthetic medical data for research and training. We are committed to medical professionalism, refining complex medical terminology to enhance model accuracy.
Data Management
We offer comprehensive data services tailored for training models, including:
1.Data Cleaning:
- Missing Value Handling: Imputation using mean, median, mode, or advanced methods like K-Nearest Neighbors (K-NN).
- Outlier Treatment: Identification and handling using Z-score, IQR, etc.
2.Data Transformation:
- Normalization: Scaling data to a [0,1] range.
- Standardization: Converting data to a distribution with a mean of 0 and standard deviation of 1.
- Log Transformation: Addressing skewed data.
3.Data Encoding:
- One-Hot Encoding: Converting categorical variables to numerical form.
- Label Encoding: Assigning integers to categorical variables.
4.Feature Selection:
Innovation, Change, Leadership
Grace AI Technology is not just a company focused on the development of generative AI application services; it is a vibrant and innovative team. We are dedicated to identifying unmet needs in the digital healthcare field and proposing innovative business models to address these challenges.
Innovation comes from God's grace.
10F., No. 452, Sec. 7, New Taipei Blvd., Xinzhuang Dist., New Taipei City 242058 , Taiwan (R.O.C.)
