What is custom vocabulary? It is a list of names and terms that you give to a recognition system so it spells them correctly. People, brands, places and technical words are typical entries. The list nudges the system toward those spellings without retraining it.
Transcribe a recording
Send your voice message, audio or video to @potok_now_bot on Telegram and the text comes back in the chat.
Open @potok_now_bot in Telegram
Files up to 2 GB · export to TXT, MD, DOCX, PDF, SRT and VTT · speaker labels and timestamps · about 30 languages
Key facts
- Rare names are the main reason for the feature, since a general model has seen few examples of them.
- Short, distinctive terms work best, and long phrases with common words add little because those words are already known.
- Not every tool offers a vocabulary list, and those that do vary in how strongly they follow it.
- Without the feature, a find-and-replace pass over the finished transcript fixes recurring names.
- Spelling variants of one name should be listed once in the form you want, so the document stays consistent from start to end.
What it looks like
// vocabulary list, one entry per line
Okafor
Kowalczyk
PostgreSQL
Kubernetes
Zhang Wei
// without such a list the same names might appear as
// "oh Kafor", "Koval check", "post grass Q L"
More
See the full list of languages, the tools and the glossary.
Frequently asked questions
What should I put into a custom vocabulary list?
Add personal names, company and product names, place names and field-specific terms. Skip everyday words, which the system already knows well.
Does the list always give the right spelling?
No. It raises the chance that a term is written your way, but the system can still pick a similar-sounding word, so a review pass remains useful.
What can I do if a tool has no vocabulary option?
Keep a list of correct spellings and run find-and-replace on the exported transcript. Do that once per recurring name rather than line by line.
Does this matter for short voice notes?
Less so. Brief notes carry few rare terms. It matters for long interviews and meetings that repeat the same unusual names many times.
How do I fix names in a Potok bot transcript?
Export it as DOCX or TXT and correct the names with find-and-replace. Check the first mention of each name against the audio.
Sources
Each source was opened by hand on the date shown. Pages change, so open the source if a detail matters to you.
- Speech recognition (Wikipedia), Wikipedia · checked
- Transcription (linguistics) (Wikipedia), Wikipedia · checked
Try it on your own recording
Send your voice message, audio or video to @potok_now_bot on Telegram and the text comes back in the chat.
Open @potok_now_bot in Telegram