Startup Builds AI System To Transcribe Meetings

Voicea, a San Francisco Bay Area startup, recently announced $20 million funding for their GPU-based deep learning system that can now fully transcribe meetings…

NVIDIA logo
NVIDIA
•
Nefi Alarcon
•2 min read•beginner•
--
•View Original
Google CloudTensorFlow

Overview

Voicea, a startup from the San Francisco Bay Area, has developed an AI system that utilizes deep learning to transcribe meetings and generate highlights. The system, supported by $20 million in funding, aims to enhance collaboration in enterprise environments by providing accurate transcriptions and integrating with popular collaboration tools.

What You'll Learn

1

How to utilize AI for meeting transcriptions and highlights

2

Why deep learning is effective for voice recognition tasks

3

When to implement AI systems for improving team collaboration

Prerequisites & Requirements

  • Understanding of AI and deep learning concepts
  • Familiarity with collaboration tools like Slack and Salesforce(optional)

Key Questions Answered

How does Voicea's AI system transcribe meetings?
Voicea's AI system uses a combination of machine learning, voice recognition, and natural language processing to join meetings and provide accurate transcriptions. It can handle multiple speakers and offers features like meeting recordings and highlights, which can be shared through platforms like Slack and Salesforce.
What technologies are used in Voicea's AI system?
The AI system is built using a local cluster of eight Quadro P4000 GPUs and the Torch deep learning framework for training neural networks. For inference, it utilizes GPUs on Google Cloud with TensorFlow, ensuring high performance and accuracy.
What funding has Voicea received for their AI system?
Voicea announced $20 million in funding to support the development of their GPU-based deep learning system aimed at transcribing meetings and enhancing collaboration in enterprise environments.
Which meeting platforms does the AI system support?
The system is designed to work in various meeting environments, including BlueJeans, Zoom, UberConference, Cisco WebEx, and Skype, making it versatile for different enterprise needs.

Key Statistics & Figures

Funding amount
$20 million
This funding supports the development of Voicea's AI system for meeting transcription.
Number of GPUs used for training
8 Quadro P4000 GPUs
These GPUs are utilized to train the neural networks that power the AI system.

Technologies & Tools

Some links below are affiliate links. We may earn a commission if you make a purchase.

Framework
Torch
Used for training the neural networks in the AI system.
Framework
Tensorflow
Used for handling the inference of the neural networks on Google Cloud.
Hardware
Quadro P4000
GPUs used for training the AI system.
Cloud Service
Google Cloud
Provides the infrastructure for running the AI system's inference.

Key Actionable Insights

1
Implementing AI for meeting transcriptions can significantly improve team productivity by reducing the time spent on note-taking.
By automating the transcription process, teams can focus on discussions and decision-making during meetings, leading to more effective collaboration.
2
Leveraging deep learning frameworks like Torch and TensorFlow can enhance the accuracy of voice recognition systems.
These frameworks provide powerful tools for training neural networks, which are essential for processing complex audio inputs in real-time.
3
Integrating AI transcription services with collaboration tools like Slack can streamline communication and information sharing.
This integration allows for easy access to meeting highlights and recordings, ensuring that all team members are aligned and informed.

Common Pitfalls

1
Failing to accurately capture multiple speakers can lead to confusion in meeting transcriptions.
This issue often arises in environments where participants speak over each other. Ensuring the AI system is trained to handle such scenarios is crucial for maintaining transcription accuracy.