Bulk replacement words in Transcription and Reprocess Summary
Don Ling
As accurate as it can get, sometimes there are accent or pronunciation that some speaker say that causes incorrect words capture. I highly recommend the feature to bulk edit words (replace).
And then an option for us to reprocess after the edit on transcription.
Example:
(Detected - Correct word)
LBO - LBU
OnCore - Angkor
Aunty - Andy
Etc...
Feel free to reach out if you need mor examples.
Autopilot
Merged in a post:
CRITICAL FEATURE: Allow Prompts to Summary to correct transcription / Summary
Don Ling
There are often mistakes in pronunciation, transcriptions and even during summary.
- Please allow us to prompt of transcript while retaining the original (i.e. a separate window / tabfl for the edits).
- Allow Prompts directly on replace words in mass or wrong interpretation of certain sentences.
It might look like...
Edit the following words in the summary while retaining every other texts.
EGO to Eagle
Idol to Idle
Baribas to Paribas
Do not make any other changes, otherwise you will be fined with $1,000,000.
Please let us fix it. Properly.
Julian Pscheid
We added a "Find" feature in the transcript edit screen. Building out the bulk replace functionality turned out to be tricky, so we're tabling that for a future addition.
Julian Pscheid
Merged in a post:
Edit Transcript Screen: Add Search or Search/Replace
Don Ling
I'm so glad that we can finally edit and rerun the transcripts now.
I faced a challenge while doing it. it's a missing search function! for example, when I found a word spoken many times throughout a session with a mistake, it was really hard for me to make changes throughout the transcripts. is there a way to bulk edit or a search and replace function?
U
Uri Martinich
I have the same issue, bulk editing (search/replace) would be incredibly useful.
Sridhar
It seems that my Quick Edit suggestion was merged into here. Please note that my suggestion goes much further than a simple search-and-replace.
Don Ling
Sridhar it's a great idea. But I wonder how long the deployment will take. I would like it as an end result and if it's quick. However I am okay with it being a search and replace if it's a quick deployment for now.
Agree that this maybe shouldn't be merged as one, since it's a different functionality
Autopilot
Merged in a post:
Quick Edit: One-stop correction for names and key terms
Sridhar
Ever noticed how AI transcription gets most words right, but struggles with the ones that matter most? When "Alex" becomes "Alec" or your product "DataSync" appears as "data sink," it creates real confusion about who's responsible for what.
The Idea
Have Hedy identify all names, proper nouns, terms, abbreviations, and terminology that appear in transcripts, then let me quickly review and correct them all in one simple interface - rather than hunting through the entire document.
How It Could Work
After transcription, Hedy would:
- Automatically flag potential names, terms, and specialized vocabulary
- Identify inconsistencies (same term transcribed differently)
- Show uncertain words it had low confidence transcribing
- Present all these terms in a single "Quick Edit" panel
- Allow me to correct once and apply throughout the document
- Learn from my corrections for future meetings
For example, after a meeting, I might see:
- "Sara/Sarah Johnson" (appears 7 times) → I select "Sarah" once
- "nexus seven/Nexus7/Nexus 7" (appears 12 times) → I select "Nexus7" once
- "A.I./AI/ai" (appears 9 times) → I select "AI" once
One quick review, and the entire transcript is instantly corrected with consistent terminology.
Why This Matters
- Save Time:Fix critical errors in seconds instead of minutes
- Ensure Accuracy:Where it matters most - names and key terms
- Improve Clarity:Consistent terminology throughout documents
- Reduce Frustration:No more hunting for errors across long transcripts
- Better Decisions:Clear task assignments and product references
This simple addition would dramatically improve the usability of Hedy's transcripts with minimal effort from users.