You're Pronouncing Words Wrong — And That's Why You Can't Understand Native Speakers

MrD — Wed, 11 Feb 2026 04:42:58 +0000

Here's a frustrating scenario every language learner knows:

You study a word. You memorize it. You use it in a sentence. But when a native speaker says it at normal speed — you don't recognize it.

Why? Because you learned the word with the wrong pronunciation. Your brain stored a wrong sound model, so when the real sound arrives, it doesn't match.

Research confirms this: pronunciation accuracy and listening comprehension are directly linked. If you can't produce the difference between "think" /θɪŋk/ and "sink" /sɪŋk/, you can't reliably hear it either.

How We Solved This in FluentCap

FluentCap is a free desktop app for real-time transcription & translation. We recently shipped two features specifically to fix the pronunciation-listening gap:

1. IPA Lookup (Offline, 25+ Languages)

Select any word from your live transcript → the International Phonetic Alphabet transcription appears instantly.

Fully offline — bundled dictionaries, no API calls
Lazy-loaded per language to keep the app lightweight
Shows exact phonetic symbols, not approximations

Example: Select "together" → /tə'ɡɛðɚ/ appears at the top of the context menu.

2. Speak (Text-to-Speech)

Right below the IPA, there's a Speak button. One click and you hear the word pronounced correctly using the Web Speech API:

const utterance = new SpeechSynthesisUtterance(selectedText);
utterance.lang = detectedLanguage;
utterance.rate = 0.9; // Slightly slower for clarity
window.speechSynthesis.speak(utterance);

The workflow: Listen → see a word → select it → read IPA → click Speak → shadow the pronunciation. This creates a feedback loop that textbooks can't provide.

Works across 25+ languages — English, Japanese, Korean, French, German, Vietnamese, Arabic, Chinese, and more.

Why This Matters

The pronunciation → listening → understanding chain is real. Break the first link and everything downstream breaks. These two features close that gap by giving you both visual (IPA) and auditory (Speak) feedback in the flow of real content.

📖 Full deep-dive article: How IPA and Speak Help You Master Pronunciation in Any Language

🌐 Try FluentCap: https://fluentcap.live — Free, no subscription, available on macOS, Windows, and Linux.

Built to bring good things to the world.

I Built a Free Desktop App for Real-time Transcription & Translation - Here's Everything It Can Do

MrD — Thu, 05 Feb 2026 04:49:03 +0000

FluentCap: A Free Desktop App for Real-time Transcription & Translation

Hi dev community!

I'm excited to share FluentCap - a desktop application I've been building to bring real-time transcription and translation to everyone. The core philosophy is simple: language should never be a barrier to understanding, learning, or connecting with others.

Check it out: https://fluentcap.live

Download: https://fluentcap.live/download

Who Is FluentCap For?

FluentCap is designed for anyone who works with audio in a different language:

Use Case	How FluentCap Helps
Language Learners	Watch K-dramas, anime, foreign films with real-time bilingual subtitles
Remote Workers	Participate confidently in international meetings
Deaf & Hard of Hearing	Access real-time captions for any audio on your computer
Researchers & Students	Transcribe lectures, podcasts, interviews instantly
Content Creators	Generate transcripts for videos without expensive services

Accessibility: Built for the Deaf & Hard of Hearing Community

One of FluentCap's core missions is accessibility. According to WHO, 433 million people worldwide experience disabling hearing loss, projected to reach 2.5 billion by 2050.

How FluentCap helps:

Real-time captions for ANY audio - not just specific apps like Zoom or Meet
Works with system audio - Netflix, YouTube, podcasts, lectures, video calls
Bilingual display - see both original language and translation simultaneously
Local processing - your audio stays on your machine, privacy-first

Research shows real-time transcription increases engagement for deaf students by 70-85%, and 98.6% of students find captions helpful for focus and comprehension.

FluentCap aims to make professional-grade captioning accessible to everyone at a fraction of the cost of traditional services (which can run $150+/hour for professional captioners). With FluentCap's BYOK model, the same utility costs just $0.15-0.40/hour.

The BYOK (Bring Your Own Key) Philosophy

Unlike subscription apps that charge $15-30/month, FluentCap uses a transparent BYOK model:

FluentCap is free forever - no ads, no subscriptions
You connect directly to speech-to-text providers using your own API keys
Providers offer hundreds of hours free to get started:

Provider	Free Credits	~Free Hours
Deepgram	$200	~750 hours
AssemblyAI	$50	~140 hours
Gladia	10 hrs/month	Unlimited (monthly reset)
Shunya	$100	~300 hours

When free credits run out, you pay providers directly at wholesale rates: $0.15-0.40/hour (60-80% cheaper than subscription apps).

Feature Highlights

1. Audio Recording & Synchronized Playback

One of the newest features! FluentCap can:

Record audio during transcription sessions (MP3 format)
Merge recording bursts into a single continuous file
Play back with full synchronization to the transcript

The 5-Pillar Interaction Model:

Automatic Recording - Audio captured as MP3 in real-time
Integrated Playback - Play/Pause, Seek, Download controls
Click-to-Seek - Click any transcript line and audio jumps there
Jump Navigation - Next/Previous highlight buttons
Bidirectional Sync - Audio playback auto-highlights & scrolls transcript

2. Highlight & Annotation System

FluentCap isn't just for passive viewing - it's an active learning tool:

Select any text - Instant highlight via context menu

Live Auto-Highlight Mode - Highlights apply immediately during recording (no menu clicks needed!)
Easy removal - Hovering over any highlight reveals a remove button

Cross-session Gallery - View all highlights from all sessions in one sidebar tab
Click-to-Navigate - Jump directly from highlight card to the exact position in transcript + audio

3. Bidirectional Sync

When audio plays:

The current segment is highlighted with a subtle flash animation
Both source & translated lists auto-scroll to keep the active segment centered
Clicking any transcript line seeks audio without changing play/pause state

4. Multiple Layout Modes

Mode	Description	Best For
Movie Mode	Large text, transparent overlay	Watching films
Meeting Notes	Visible sidebar, high opacity	Work calls
Side-by-Side	Vertical comparison view	Language learning

5. Multi-Source Audio Capture

Audio Source	Use Case
System Audio	Netflix, YouTube, Podcasts
Microphone	Dictation, Note-taking
Both	Video calls (hear everyone)

6. Customizable Themes

FluentCap offers multiple themes to match your preference and reduce eye strain during long sessions.

Privacy-First Architecture

Local storage only - all sessions saved on your machine
Direct API calls - audio goes straight to providers, no intermediate servers
No tracking - your data stays yours

Technical Stack

For the devs interested in the architecture:

Framework: Electron + React + TypeScript
Audio Processing: FFmpeg (via ffmpeg-static)
Storage: JSON-based local persistence
Virtualization: react-window for smooth transcript scrolling
STT Providers: Deepgram, AssemblyAI, Gladia, Shunya

Thank You to Our Providers

FluentCap exists because of the amazing speech-to-text providers who make their technology accessible:

Deepgram - Lightning-fast Nova-3 model
AssemblyAI - Excellent accuracy and features
Gladia - Generous monthly free tier
Shunya - Budget-friendly rates

When your free credits run out, please consider supporting them. Their pricing is incredibly fair — just $0.15-0.40 per hour, and they deserve recognition for democratizing this technology.

Try It Out

Website: https://fluentcap.live

Download: https://fluentcap.live/download

Blog: https://fluentcap.live/blog

Available for macOS, Windows, and Linux.

Looking for Feedback

I'd love to hear from the community:

What features would make this more useful for your workflow?
Any accessibility improvements you'd suggest?
Feedback on the BYOK model?

Drop a comment below or reach out through the website. Built to bring good things to the world.

— FluentCap Team

A world without language barriers.

DEV Community: MrD