More Customization, Accuracy Gains & Insights: Explore VoiceAI’s New Features

Felix Laumann
NeuralSpace
Published in
3 min readJan 25, 2024

We’re releasing a new version of VoiceAI and it’s packed with fresh features!

With generative AI, enhanced multichannel diarization, custom vocabulary, subtitle guidelines, and improved language detection, it’s more than an upgrade — it’s a whole new experience.

Try VoiceAI for free at voice.neuralspace.ai — your solution for accurate transcription, in-depth analysis, and interactive conversations.

Ask me Anything: Turn Audio into Insights

What happens when generative AI meets VoiceAI? Introducing Ask me Anything.

Extracting key insights from transcriptions has always been a challenge. “Ask me Anything” changes that. Simply type a question and instantly receive answers.

No more sifting through data — just direct, instant access to the information you need. Rest assured that when you use Ask me Anything, you can only search and retrieve information from your own audio transcripts.

Here’s some prompt inspiration to help you get started:

  • Fraud detection: “Did this caller fail multiple attempts to access their account?”
  • Sentiment by topic: “Summarize how this caller felt about their purchase and why.”
  • Interview: “What are the candidates’ strengths and weaknesses?”
  • Content creation: “Create a description of this podcast episode in less than 500 words.”

Custom Vocabulary: Tailor Your Model

Unique language and terminology can often go unrecognised or wrongly transcribed by standard Speech-to-Text (STT) systems. Bad transcriptions can disrupt your workflow, leading to distorted speech analytics and faltering CX systems.

With VoiceAI’s Custom Vocabulary feature, you can add any word — from product names to specialist industry lingo — and see instant improvements in transcription accuracy. Achieve precise results without the need for complex model training.

Multichannel Diarization: Clarity in Audio Segmentation

Distinguishing between different speakers in a single audio stream can be challenging, especially if the speakers’ voices are similar or if there’s background noise. Multichannel diarization solves this by processing each speaker’s audio from separate channels. This clear separation allows the system to more accurately identify who said what and when.

VoiceAI now offers two diarization modes:

Speaker Mode: Identify different voices within one audio channel.

Channel Mode: Use when each speaker’s audio is separated into different channels.

Advanced Configuration for Diarization

Enhance your speaker identification accuracy with advanced diarization settings. Choose from the pre-set configuration or adjust the sensitivity slider for perfect speaker separation.

High sensitivity detects more speakers; low sensitivity, fewer. Know your speaker count? Input it directly to ensure the most accurate transcription results.

In this release, we also introduce subtitle guidelines — a feature that allows you to tailor text length, line count, and duration. Easily apply your preferences and download an SRT file that reflects your personalized settings — perfect for content creators. We’ve also upgraded our language detection model, enhancing the accuracy of your transcriptions.

Subtitle Guidelines

Sign up to the VoieAI platform to try these features for free.

Contact our sales team with any questions about our enterprise pricing and bespoke solutions. We’re here to help.

--

--