LSTN - AI Transcription SaaS

Overview

Led development of LSTN, a GPT-powered B2B transcription tool serving major research companies Ipsos and Nielsen. Built as part of Supry Inc's product suite (2021-2022).

The Challenge

Traditional transcription services are slow, expensive, and error-prone:

• •Manual transcription takes hours for each hour of audio
• •Accuracy varies significantly between transcribers
• •No real-time processing capabilities
• •Poor integration with research workflows

The Solution

LSTN revolutionizes transcription through AI and real-time processing:

1. GPT-Powered Transcription

• •Leveraged GPT for context-aware transcription
• •Improved accuracy by 40% over traditional methods
• •Handled industry-specific terminology

2. Real-time Chat Interface

• •Built chat-based transcription system
• •Users interact with transcription in real-time
• •Immediate corrections and edits
• •Natural conversation flow

3. B2B Platform

• •Designed for research companies (Ipsos, Nielsen)
• •Batch processing for large volumes
• •Team collaboration features
• •Export in multiple formats

Technical Implementation

• •GPT integration for transcription
• •Python backend for audio processing
• •Real-time WebSocket connections
• •Firebase for data storage

• •Next.js for web application
• •Real-time chat interface
• •Audio player integration
• •Collaborative editing

• •Firebase Realtime Database
• •Cloud storage for audio files
• •CDN for fast delivery
• •Scalable processing pipeline

Key Features

• •Real-time Transcription - See results as you speak
• •Chat Interface - Natural interaction model
• •High Accuracy - 40% better than competitors
• •B2B Ready - Designed for enterprise workflows
• •Multi-format Export - SRT, VTT, TXT, DOCX

Impact

• •40% accuracy improvement over traditional transcription
• •Ipsos & Nielsen as major clients
• •Real-time processing enabling new workflows
• •Scalable platform handling large volumes

Lessons Learned

• •GPT can significantly improve transcription accuracy
• •Chat-based interfaces feel more natural for corrections
• •Real-time feedback is critical for user experience
• •B2B requires robust batch processing and export features

---

*Supry Inc project: AI transcription platform, 2021-2022*