This is a submission for the AssemblyAI Challenge : Sophisticated Speech-to-Text.
What I Built
I developed a tool designed to help sales professionals improve their phone call performance by offering AI-powered analysis and practice. The tool uses AssemblyAI’s to transcribe and analyze sales calls, providing actionable feedback to users on their tone, language, pacing, and key performance indicators (KPIs) such as engagement and call effectiveness.
The application is designed to help salespeople identify areas for improvement, whether it's making their pitch more compelling, asking the right questions, or learning how to better respond to customer objections. By using AI to transcribe and analyze their calls, salespeople can practice and refine their techniques to increase their chances of closing deals.
Demo
You can access a live demo of the tool here. Below are some screenshots showcasing the key features of the application:
You can checkout the code here
Call Transcription Screen: This is where the user can upload their sales call recording. The AI will then transcribe the call and display the text for review.
Analysis Dashboard: After transcription, the AI analyzes the call and provides a detailed breakdown of language use, tone, and response effectiveness.
Practice Mode: Users can practice their pitch by simulating conversations with the AI, receiving real-time feedback on their responses.
Journey
After seeing the dev.to challenge, I wanted to build something useful. Initially, I thought about creating a platform to help people learn English with the aid of AI. However, I decided to shift focus and build something more practical for salespeople. As a result, I developed this platform and made it available for free to everyone.
Step-by-Step Implementation
Transcription with AssemblyAI: The first step was to integrate AssemblyAI’s API to convert sales call audio recordings into text.
Call Analysis: Once the call was transcribed, I used OpenAI or Ollama model based on user configured on setting page.
Real-time Feedback: The user can choose the type of prospect they want to engage with, such as positive, negative, busy, etc. Once they select the prospect persona, they can start speaking, and the audio will be transcribed to text using AssemblyAI. The transcription will then be sent to the AI with a prompt, and the AI's response will be displayed on the UI. The user can continue the conversation, and once they’re done, they can analyze the call to see how well they performed.
Finally, all the best to everyone participating in this hackathon!
Happy Coding :)
Top comments (1)
Thanks for your sharing