How to Add Custom Vocabulary for Technical Terms in Sonicribe
Master Sonicribe's custom vocabulary feature to perfectly transcribe technical jargon, brand names, programming terms, and industry-specific language every time.
Sonicribe Team
Product Team
Table of Contents
Why Custom Vocabulary Matters
Ever dictated "Kubernetes" and gotten "Cooper Netties"? Or said "GraphQL" only to see "Graphic Wool"? Standard speech-to-text struggles with:
- Technical jargon (API, SDK, OAuth, CI/CD)
- Brand names (PostgreSQL, FastAPI, Tailwind)
- Made-up words (your product names, character names)
- Acronyms specific to your industry
- Names of people you mention frequently
Sonicribe's custom vocabulary fixes this permanently. Add a word once, and it's recognized correctly forever—with zero cloud dependency.
How Custom Vocabulary Works
Unlike cloud services that "learn" from your usage (by storing your data), Sonicribe's vocabulary system works differently:
1. You define words explicitly — Full control over what's recognized
2. Processing stays local — Your vocabulary never leaves your Mac
3. Instant recognition — No training period required
4. Smart replacements — Transform spoken phrases into formatted output
Adding Your First Custom Word
Step 1: Open Vocabulary Settings
1. Click the Sonicribe icon in your menubar
2. Select Settings (gear icon)
3. Navigate to the Vocabulary tab
Or use the keyboard shortcut: Cmd + , then click Vocabulary.
Step 2: Add a Word
Click the + Add Word button, then fill in:
- Spoken form: How you'll say the word ("kubernetes")
- Written form: How it should appear ("Kubernetes")
- Category (optional): For organization ("DevOps", "Programming")
Click Save and you're done.
Step 3: Test It
Press your transcription hotkey (default: Alt+Space), say "I deployed to Kubernetes," and watch the magic happen.
Read more: Custom Vocabulary for Legal Terms: Court, Contracts & Compliance
Essential Vocabulary for Developers
Here are the most commonly needed technical terms. You can add these manually or import our pre-built Developer Pack:
Programming Languages & Frameworks
| Spoken | Written | Notes |
|---|---|---|
| "react" | React | Capitalized framework name |
| "next js" | Next.js | With correct styling |
| "typescript" | TypeScript | Proper casing |
| "javascript" | JavaScript | Proper casing |
| "python" | Python | Capitalized |
| "node js" | Node.js | Styled correctly |
| "tailwind" | Tailwind CSS | Full name |
| "postgres" | PostgreSQL | Full name |
| "mongo db" | MongoDB | Correct casing |
Developer Tools
| Spoken | Written |
|---|---|
| "git hub" | GitHub |
| "git lab" | GitLab |
| "docker" | Docker |
| "kubernetes" | Kubernetes |
| "v s code" | VS Code |
| "x code" | Xcode |
| "homebrew" | Homebrew |
| "npm" | npm |
| "yarn" | Yarn |
Common Acronyms
| Spoken | Written |
|---|---|
| "api" | API |
| "sdk" | SDK |
| "rest api" | REST API |
| "graphql" | GraphQL |
| "oauth" | OAuth |
| "jwt" | JWT |
| "ci cd" | CI/CD |
| "aws" | AWS |
| "gcp" | GCP |
Smart Replacements: Beyond Single Words
The real power of custom vocabulary is smart replacements—transforming spoken phrases into formatted output:
URLs and Links
| Spoken | Written |
|---|---|
| "my calendar link" | https://cal.com/yourname |
| "company website" | https://yourcompany.com |
| "github profile" | https://github.com/username |
| "linkedin" | https://linkedin.com/in/yourname |
Code Snippets
| Spoken | Written |
|---|---|
| "console log" | console.log() |
| "use state hook" | useState() |
| "use effect hook" | useEffect() |
| "async function" | async function |
| "arrow function" | () => {} |
Email Signatures
| Spoken | Written |
|---|---|
| "my email signature" | Best regards,\nJohn Smith\nSenior Developer |
| "quick sign off" | — John |
Common Phrases
| Spoken | Written |
|---|---|
| "let's hop on a call" | Let's schedule a call to discuss. |
| "LGTM" | Looks good to me! |
| "will do" | I'll take care of that. |
Using Pre-Built Vocabulary Packs
Sonicribe includes pre-built vocabulary packs for common industries:
Available Packs:
1. Software Development — 200+ programming terms
2. Medical/Healthcare — HIPAA-compliant terminology
3. Legal — Court terms, case law references
4. Finance — Accounting, trading terminology
5. Sales — CRM terms, sales methodology
6. Construction — Building, safety terminology
7. Logistics — Supply chain, shipping terms
Installing a Pack:
1. Go to Settings → Vocabulary
2. Click Browse Packs
3. Select the pack you need
4. Click Install
Packs don't overwrite your custom entries—they add to them.
Organizing Your Vocabulary
As your vocabulary grows, organization matters:
Use Categories
Create categories that match your work:
Read more: Custom Vocabulary for Programming: APIs, Frameworks & Libraries
- Company — Brand names, product names, colleague names
- Project — Current project-specific terms
- Clients — Client names and their product terms
- Code — Programming-specific vocabulary
Enable/Disable Categories
Working on different projects? Toggle categories on/off:
1. Go to Settings → Vocabulary
2. Click the toggle next to any category
3. Disabled categories won't affect transcription
Export and Import
Moving to a new Mac? Share vocabulary with your team?
Export:1. Settings → Vocabulary → Export
2. Save the JSON file
Import:1. Settings → Vocabulary → Import
2. Select your JSON file
3. Choose "Merge" or "Replace"
Mode-Specific Vocabulary
Sonicribe lets you assign vocabulary to specific modes:
Example: Code Mode
Create a "Code" mode that activates when you're in VS Code:
1. Settings → Modes → Create Mode
2. Name: "Code"
3. Auto-activate in: VS Code, Terminal, Xcode
4. Vocabulary: Enable only "Code" category
Read more: How Sonicribe Keeps Your Voice Data Private: Zero Cloud Architecture
Now when you dictate in VS Code, only programming vocabulary is active—no interference from medical terms or client names.
Example: Email Mode
1. Name: "Email"
2. Auto-activate in: Mail, Gmail (browser)
3. Vocabulary: "Email Signatures", "Client Names"
Best Practices
1. Start Small
Don't try to add 500 words on day one. Add terms as you encounter misrecognitions.
2. Be Specific with Spoken Forms
If Whisper misrecognizes "Terraform" as "Terra Form," add:
- Spoken: "terra form"
- Written: "Terraform"
Listen to how you naturally say words—that's your spoken form.
3. Use Phonetic Alternatives
For words with multiple pronunciations:
| Spoken | Written |
|---|---|
| "postgres" | PostgreSQL |
| "postgre sql" | PostgreSQL |
| "post gres" | PostgreSQL |
Add all variations pointing to the same output.
4. Leverage Smart Replacements
The biggest time-savers are phrase replacements:
Read more: Sonicribe vs Descript: Dictation vs Content Editing
- "my zoom link" → Your actual Zoom URL
- "my phone number" → Your formatted phone number
- "standard disclaimer" → Your legal boilerplate
5. Review and Prune
Monthly, review your vocabulary:
- Remove terms you no longer use
- Update outdated links/information
- Consolidate duplicate entries
Troubleshooting
Word Not Being Recognized
- Check spelling of spoken form (lowercase, as you say it)
- Try adding phonetic alternatives
- Verify the category is enabled
Replacement Not Working
- Ensure exact spoken form match
- Check for spaces (vs. no spaces) in multi-word phrases
- Verify the entry isn't disabled
Vocabulary Slowing Down Transcription
This shouldn't happen with normal vocabulary sizes (under 1000 words). If it does:
1. Export your vocabulary
2. Check for duplicates
3. Remove unused entries
The Power of Personalized Transcription
Custom vocabulary transforms Sonicribe from a general transcription tool into your personal dictation assistant. The more you customize, the faster and more accurate your workflow becomes.
Technical terms that once required manual correction now transcribe perfectly. URLs you type daily now flow from speech. And all of it happens locally, on your Mac, with zero cloud involvement.
Your vocabulary is your data. Keep it private, keep it powerful.
Ready to build your perfect vocabulary? Download Sonicribe and start customizing today.
Related Reading
Ready to transform your workflow?
Join thousands of professionals using Sonicribe for fast, private, offline transcription.
