Corti’s speech recognition technology is built specifically for the healthcare sector, helping professionals convert speech into text accurately and efficiently. This guide explains what Corti speech recognition can do, the available options (called “endpoints”), and how to choose the best one for your needs.
What is Corti Speech Recognition?
Corti speech recognition is designed to turn spoken language into written text in real time or after recording. It helps with:
Dictation (writing as you speak)
Medical documentation
Decision support during conversations
It is tailored for healthcare use, offering high accuracy, medical term recognition, and flexibility.
👉 If you want to know which languages are supported, please check our Languages page for details on languages, features, and codes you need for API use.
What options (endpoints) can I use?
Corti offers three types of speech recognition services depending on your workflow:
Endpoint | Best for | How it works |
Transcribe | Dictation (typing as you speak) | Works in real time, no memory of earlier speech |
Stream | Live medical conversations and decision support | Provides live transcripts and identifies key facts |
Transcripts | Processing audio files you upload | Converts recorded audio to text after the recording is done |
Key Features by Endpoint
Feature | Transcribe | Stream | Transcripts |
Connection type | Real-time connection | Real-time connection | Upload files |
How it processes speech | Immediate results | Immediate results + insights | Processes after upload |
Memory of conversation | No (stateless) | Yes (remembers context) | Yes (remembers context) |
Type of text | Verbatim (word for word) | Conversational summary | Verbatim or conversational summary |
Speaker separation (diarization) | ❌ | Optional | Optional |
Multichannel support | ❌ | Optional | Optional |
Custom commands (e.g., “insert template”) | ✅ | ❌ | ❌ |
Automatic punctuation | Optional | ✅ | Optional |
Spoken punctuation (e.g., “comma”) | Optional | ❌ | Optional |
📝 We’re working on adding smart formatting and custom dictionary options soon!
How Corti Speech Recognition Works
Corti uses advanced speech recognition technology that combines different types of models:
One model focuses on understanding the sound of speech
Another model focuses on understanding the meaning of words
Together, these models help:
Reduce mistakes
Make speech-to-text faster
Balance accuracy and speed based on your needs
How does Corti improve its models?
Corti continuously works to make its speech recognition better by:
Training on specialized healthcare language and terms
Testing with different data than what’s used for training
Measuring quality using:
Word error rate
Medical term accuracy
How close the text is to what was said (Levenshtein distance)