![case study](https://cdn.quantiphi.com/2024/02/0b844f1f-specialty-speech-recognition.png)
Business Impacts
100%
Accuracy for number transcriptions
95%
Accuracy for trading parameter transcriptions
Customer Key Facts
- Location : North America
- Industry : Technology
Challenges
- Overlapping speech data
- Transcription of financial jargons
- Lack of labeled data
- 5-10 seconds of trade “shouts” in multiple accents
![](https://cdn.quantiphi.com/2024/02/Q_Specialty-SR-For-Financial-Day-challenges.png)
Technologies Used
![TensorFlow](https://cdn.quantiphi.com/2024/02/tensor-flow.png)
TensorFlow
![Python](https://cdn.quantiphi.com/2024/02/5-2-python-logo-png-image-thumb-1.png)
Python
![Google ML Engine](https://cdn.quantiphi.com/2024/02/ML-engine.png)
Google ML Engine
Monitoring Financial Trader Behavior with a Domain-Adapted Speech Recognition Model
Solution
Quantiphi built a state-of-the-art speech recognition model for monitoring financial trader behavior. The custom Automated Speech Recognition (ASR) helped to accurately transcribe trader conversations with financial jargon and domain-specific terms to text.
Cloud9 Technologies also uses Quantiphi to inject powerful analytics into its platform driven by the data collected, providing users with the vital information they need on the trading floor. This includes text analytics-driven insights for frequently traded commodities, active buyers, purchased stocks, and more, further enhancing relationships with traders.
Results
- Better monitoring of trader behavior through domain-adapted custom speech recognition
- Speech transcriptions with near human level accuracy
- Streamlined and simplified storage of exceptionally large data volumes