AI-powered healthcare assistant enhances medical transcription by 120% with Gladia
Published on Feb 28, 2025
Medical transcription is among the most critical and challenging verticals for ASR models to date.
Filled with drug names and medical jargon, medical consultations, dictations, and online conferences require versatile solutions, with custom vocabulary and specialized models needed to make speech-to-text solutions attuned to jargon. There’s the issue of security too, as audio from medical consultations is among the most sensitive confidential data out there.
A fast-growing healthcare generative AI startup, who prefers to remain anonymous, turned to Gladia for top-quality medical transcription at scale. Here’s how we helped them increase their accuracy and speed of transcription, all while ensuring 100% security of confidential user data.
Challenge
Doctors spend about 60% of their time on computers, doing non-clinical work. This startup is aiming to get that number to 15%, enabling doctors to allocate most of their time for consultation, diagnostics, and other high-value tasks with the help of AI.
They knew that having accurate transcription for note-taking during consultations was the first step in designing a holistic solution to achieve this milestone.
Indeed, the platform’s ability to understand and actively transcribe jargon-filled medical conversations is an essential prerequisite for LLM-powered notes, prescriptions, and intricate EHR enrichment that distinguish their AI co-pilot.
Speed is likewise a key factor for them, as the ability to generate notes shortly after the consultation is critical for efficient clinical workflows.
Moreover, they needed to ensure 100% protection of all user data in accordance with HIPAA and GDPR, which most of the US-based providers are generally not able to provide.
This is why their team took the task of choosing a speech-to-text provider very seriously. With regular evaluations in place, they have tested over 7 different providers before, including the Big Tech cloud solutions — all of which ultimately failed to strike the right balance between accuracy, speed, price, and security standards.
Solution
With Gladia, the team was able to implement:
Highly accurate transcription solution, registering a Word Accuracy Rate between 95% and 98%, with near-human-level performance in English;
Near real-time speed of batch transcription;
Custom vocabulary, which ensures correct transcription of drug names and other medical jargon;
On-premisehosting, with quick setup and 24/7 dedicated engineering support;
Certification in compliance with HIPAA and GDPR.
Impact
Following a swift onboarding and a dedicated on-premise environment setup, they began to use Gladia as its primary speech-to-text provider. The results did not take long to show.
By working with the Gladia team to iterate and scale up, they saw a noticeable impact on their system’s performance:
The team was likewise impressed by the quality of Gladia’s technical assistance, allowing them to not only set up their dedicated environment in a matter of hours but also benefit from Gladia’s in-house engineering expertise to optimize their infrastructure as a whole.
Given the initial success with Gladia API and its on-premise deployment, this innovative company is already considering how they will leverage our product in the future as they extend their platform to new stakeholders.
For instance, they look forward to experimenting more with multilingual transcription and translation, which would enable patients to consult physicians in their native language. They also intend to leverage speaker diarization for collective medical meetings.
Having read this case study, do you feel like Gladia API can be the right fit for your business?
STT API Benchmarks: How to measure accuracy, latency, and real-world Performance
Every product that depends on voice input lives or dies by its speech-to-text performance. Whether you're enriching CRM data from support calls, powering live captions in meetings, or triggering downstream actions via LLMs, transcription accuracy and speed aren’t just nice-to-haves. They’re essential to product functionality. If your STT engine stalls on latency or mistranscribes a customer’s request, it can break automations, derail user experiences, and create costly manual work downstream.
As the landscape of speech-to-text APIs continues to evolve—with growing demands around latency, language support, and compliance—it’s more important than ever to ensure that your setup aligns with your product’s direction.
Gladia and Pipecat partner to push the boundaries of real-time voice AI
We’re thrilled to announce a strategic partnership between Gladia and Daily, the team behind Pipecat, aimed at revolutionizing real-time conversational AI. This collaboration combines our cutting-edge audio intelligence capabilities with their flexible 100% open-source framework, empowering developers to create more dynamic, multilingual, and context-aware voice AI applications.