Physicians are currently able to to record their voice notes in a floating window on Cirrus and to link them to a text transcription of this audio. This to allow them taking voice notes during their EHR reviewing process. However, the transcription of these recording had to be entered manually by either the nurse or the physician.
In order to improve this process, we have now implemented an AI speech recognition tool from OpenAI called Whisper , which is an automatic speech recognition (ASR) system trained on 680,000 hours of multilingual and multitask supervised data collected from the web. With this tool, users will be able to upload their own voice notes and to automatically generate their transcriptions in their own language (you can set as default and Whisper will recognize the language). See all details of this improvement below.
Improved Voice Notes Recording Logic
In order to always save your recorded voice notes on the dedicated server for them, we have now changed the logic of how these voice notes are saved. Now, before downloading the voice notes locally, you will require to Save them first. After saving them, you will be able to download them locally.
With this improvement, we will allow users to have these files always on their cloud storage.
1. Once you have stopped recording a voice note, click on Save to server. You will see a confirmation message after its completion, this will enable the Download to device button.
Click Download to device if you wish to save the recording locally on your PC.
Audio Transcribing
Now, with the help of OpenAI's Whisper transcriber, physicians can import their own voice note files in order to automatically get the transcription for these voice orders that will be linked to their related form. This enhancement will apply for clients that use Worklist for recording voice notes.
The settings needed for this are the following :
FTP Server configuration for recording voice notes.
Whisper Secret Key for voice transcriptions.
With this feature, users will be allowed to upload voice notes in their own language. See the step by step here.
After recording the audio, in Worklist click context menu and select Audio Transcribing :
2. A new window will open up. Click on File Attachment .
3. Now click on Browse... and select your local file.
4. Your file will appear attached, click Upload.
5. Now, select the language of this audio from the list.
6. To complete, click on Transcribe :
Tanscription will appear in the text field below. You can edit this information if required.
Demo :
Other screens availability:
The Audio Transcribing feature will also be accessible from Form Section Search, in context menu :
After clicking it, you will be able to attach (see the forementioned steps 1-6 from above) the audio files :
Upcoming Releases :
View generated transcriptions (TBD)
We hope this improves your Cirrus experience. Remember to share this information and ask for help if you need it.
April 10, 2023.