Quick Facts
- Product: DopaSpeak, an Android language learning app using on-device AI for word-by-word song decoding
- Languages supported: 20+ languages including Spanish, French, German, Italian, Portuguese, Japanese, Korean, Mandarin, Hindi, Arabic, Turkish, Russian
- AI model: Google Gemma 4 quantized, runs entirely on-device
- Pricing: Free tier, $7.99 USD per month, or $199 USD Lifetime (no auto-renewal)
- Platform: Android (API 24+, optimized for Android 11+)
- Founded: 2025 by Rene Scafarti, operated by Scafa Investments LLC
- Status: Beta with 3-person tester program, soft-launch via web (dopaspeak.com)
- Headquarters: Cutler Bay, Florida, United States
- Press contact: info@scafa-investments.com
Boilerplate Text
Three lengths for different contexts. Use as-is, no quotation marks needed.
50 words
DopaSpeak is an Android language learning app that decodes song lyrics word by word using Google Gemma on-device. Users import any music they love, the app shows the original text, literal translation, and natural translation side by side. 20+ languages, no subscription trap.
100 words
DopaSpeak is an Android language learning app built around the natural decoding method: word-by-word translation of songs the user already loves, in 20+ languages. The app uses Google Gemma 4 quantized for AI inference entirely on-device, meaning nothing leaves the phone, important for users wary of cloud services. Pricing is $7.99 per month or $199 Lifetime, with no auto-renewal. Built by indie developer Rene Scafarti and operated by Scafa Investments LLC, DopaSpeak targets markets where existing language apps are too expensive, too cloud-dependent, or too gamified to actually teach fluency.
200 words
DopaSpeak is an Android language learning app that operationalizes Stephen Krashen's Input Hypothesis through music. Instead of flashcards or streak-based gamification, users import any song they already love, and the app decodes every word using Google Gemma 4 quantized, running entirely on-device. The user sees the original lyrics, the literal word-for-word translation in their own language, and the natural translation for comparison. The brain learns from comprehensible input it is emotionally invested in, not from isolated drills.
The app supports 20+ languages, including Spanish, French, German, Italian, Portuguese, Japanese, Korean, Mandarin, Hindi, Arabic, Turkish and Russian. Pricing is $7.99 per month or $199 Lifetime, with no auto-renewal trap. The on-device AI is a deliberate choice: nothing leaves the phone, no cloud account is required, and the app works offline after the initial download.
DopaSpeak was built by indie developer Rene Scafarti and is operated by Scafa Investments LLC, headquartered in Cutler Bay, Florida. The launch focuses on markets where existing apps are too expensive (Brazil, Turkey), too cloud-dependent (privacy-conscious diaspora), or too gamified for actual fluency goals.
Founder Bio
Rene Scafarti is an indie developer based in Germany and the founder of DopaSpeak. He built the app to solve a personal frustration: spending years on flashcard apps without ever being able to follow a song lyric in a foreign language. After encountering Krashen's Input Hypothesis and realizing that on-device AI had become powerful enough to decode language in real-time, he started prototyping in 2025. DopaSpeak operates under Scafa Investments LLC, which is registered in Florida.
Beta-Tester Quotes
The app launched with a closed beta of three testers, all of whom appear with quotes on the product pages. Use these quotes verbatim if needed.
"Three months ago I knew five words of Spanish. Today I understand entire songs word by word. The decoding method finally makes language structure visible."
, Aljoscha Pampuch, beta tester
"Finally, no green owls, no streak pressure, no ads. My data stays on my phone, and the AI decodes at my actual level."
, Vivian Nemitz, beta tester
"I tried Babbel, Duolingo, Pimsleur. DopaSpeak is the first app that adapts to music I already listen to instead of forcing a curriculum."
, Ron Chiovaro, beta tester
Technical Specifications
- AI inference: Google Gemma 4, quantized for mobile, fully on-device
- Platform: Android, native Kotlin, Jetpack Compose
- Minimum API: Android 24 (Android 7.0). Optimized for Android 11 and above.
- RAM target: Works on 4 GB devices, recommended 6 GB for fastest inference
- Storage: ~2 GB after model download
- Network: Required for initial setup and song import. All learning works offline.
- Privacy: No cloud account, no telemetry, no third-party tracking. GDPR-compliant by architecture.
- Cloud APIs (optional): Users can add their own Gemini, OpenAI, Anthropic, OpenRouter or Moonshot keys via encrypted on-device storage. Default mode is fully on-device.
Pricing in 13 Markets
DopaSpeak uses local pricing per market to avoid the EUR-conversion friction that hurts Babbel and Duolingo in emerging markets.
| Market | Per Month | Lifetime |
|---|---|---|
| USA / English | $7.99 | $199 |
| Germany, Austria, Switzerland | 7,99 € | 199 € |
| Brazil | R$ 48 | R$ 1.199 |
| Turkey | 89 TL | 2.299 TL |
| Japan | ¥1.290 | ¥29.900 |
| South Korea | ₩9.990 | ₩199.000 |
| China | ¥58 | ¥1.499 |
| India | ₹599 | ₹14.999 |
| Saudi Arabia, UAE, Egypt | $7.99 | $199 |
Market-Specific Story Hooks
DopaSpeak built dedicated landing pages for five markets where the language learning need has a specific narrative angle. Each is a potential story angle for regional press.
- Saudi Arabia / Vision 2030: English for Vision 2030 careers, Saudi Arabia's national strategy explicitly identifies English proficiency as a competitive necessity. NEOM employs 100+ nationalities; 80% of the wider workforce is expat.
- Türkiye / Chancenkarte: German for the Chancenkarte, Germany's points-based work visa, active since June 2024. German-language searches in Türkiye tripled, but mainstream apps are TL-inflation-priced.
- India / K-Drama Wave: Korean drama in Hindi, 246,000 monthly searches in India. The same volume as English-app searches. No app served this audience until DopaSpeak.
- Brazil / Nikkei Heritage: Japanese for the Nikkei diaspora, São Paulo hosts the largest Japanese diaspora outside Japan, 1.5 million people across three generations.
- USA / Krashen Method: Natural Decoding Method, Stephen Krashen, the linguist behind the Input Hypothesis, was a Spanish teacher in California. DopaSpeak operationalizes his framework as a consumer app.
Downloadable Assets
Free for editorial use with attribution to "DopaSpeak / Scafa Investments LLC". Right-click and save, or contact us for full-resolution masters.
App Icon
app-icon-512.png (RGB, 512×512, transparent background)
Screenshots (Android phone, hi-res)
- screen1-decode.png, Word-by-word decoding interface
- screen2-songs.png, Imported song library
- screen3-ai-settings.png, On-device AI configuration
- screen4-progress.png, Progress and vocabulary tracking
Tablet Screenshots
- tablet10-screen1.png, Tablet view, decoding
- tablet10-screen2.png, Tablet view, library
Feature Graphic
feature-graphic-1024x500.png (Play Store header format)
Open Graph Image
og-image.png (1200×630, suitable for social-media embeds)
Press Contact
Scafa Investments LLC
9830 Bahama Dr
Cutler Bay, FL 33189-1568
United States
Email: info@scafa-investments.com
Founder direct (developer-context): j.scafarti@scafa.de
Phone: +49 170 322 0161
Web: scafa-investments.com
For interview requests, demo builds, beta-tester contact, or specific data points, reach out by email. Replies typically within 24 hours.
What Not To Write About Us
Two things we ask press to avoid:
- No "fluent in 30 days" claims. The app supports realistic progress: understanding song lyrics within 4-6 weeks, conversational level within 6-12 months, depending on language distance and time investment. Hyperbolic claims hurt credibility for everyone in the space.
- No artist or song names from the app. The user imports their own music, so any specific artist mention misrepresents what the app does. Use genres instead: K-Pop, J-Pop, Reggaeton, Bollywood, Anadolu Rock, MPB, Cantautori.
Last updated: May 2026