The latest articles on DEV Community by Tomáš Dobrý (@tubevoice). https://dev.to/tubevoice https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F3848095%2F233e8039-cc0c-4493-9866-13804e1bc480.png https://dev.to/tubevoice en Tomáš Dobrý Sat, 28 Mar 2026 18:04:16 +0000 https://dev.to/tubevoice/i-built-an-ai-tool-that-dubs-any-youtube-video-into-50-languages-3bd1 https://dev.to/tubevoice/i-built-an-ai-tool-that-dubs-any-youtube-video-into-50-languages-3bd1 <p>There are millions of great YouTube videos out there. Tutorials, podcasts, documentaries, lectures. But most of them are locked behind a language barrier.</p> <p>I built <strong>TubeVoice</strong> to fix that.</p> <h2> What it does </h2> <p>Paste any YouTube URL, pick a target language, and TubeVoice generates a fully dubbed version in minutes. The original background music and ambient sounds stay intact — only the voice changes.</p> <p><strong>Try it free:</strong> <a href="https://tubevoice.io" rel="noopener noreferrer">tubevoice.io</a></p> <h2> How it works </h2> <ol> <li> <strong>Transcription</strong> — AI transcribes the original audio (Whisper)</li> <li> <strong>Translation</strong> — Neural machine translation to your chosen language</li> <li> <strong>Voice Synthesis</strong> — Natural-sounding TTS generates the dubbed voice</li> <li> <strong>Audio Mixing</strong> — AI separates vocals from background using source separation (Demucs), then mixes the new voice with the original background audio</li> </ol> <p>The result sounds surprisingly natural. The timing is preserved, background music plays through, and the dubbed voice matches the original pacing.</p> <h2> Tech Stack </h2> <ul> <li> <strong>Frontend:</strong> Next.js + Tailwind CSS (deployed on Vercel)</li> <li> <strong>Backend:</strong> Python + Celery workers (Railway)</li> <li> <strong>Transcription:</strong> OpenAI Whisper</li> <li> <strong>Translation:</strong> GPT-4o</li> <li> <strong>TTS:</strong> Google Chirp3-HD (Basic), ElevenLabs (Standard/Premium)</li> <li> <strong>Audio Separation:</strong> Demucs (htdemucs model)</li> <li> <strong>Database:</strong> Supabase (PostgreSQL)</li> <li> <strong>Payments:</strong> Stripe</li> </ul> <h2> Three quality tiers </h2> <div class="table-wrapper-paragraph"><table> <thead> <tr> <th>Tier</th> <th>Voice Engine</th> <th>Credits/min</th> <th>Best for</th> </tr> </thead> <tbody> <tr> <td>Basic</td> <td>Google Chirp3-HD</td> <td>1</td> <td>Casual listening</td> </tr> <tr> <td>Standard</td> <td>ElevenLabs Flash</td> <td>3</td> <td>Good quality dubbing</td> </tr> <tr> <td>Premium</td> <td>ElevenLabs Dubbing API</td> <td>6</td> <td>Professional results</td> </tr> </tbody> </table></div> <h2> Challenges I faced </h2> <ul> <li> <strong>Audio sync</strong> — Matching dubbed speech timing to the original is hard. Different languages have different word lengths. The TTS engine handles this reasonably well, but it's not perfect.</li> <li> <strong>Background preservation</strong> — Using Demucs for source separation was a game-changer. It cleanly separates vocals from music/ambience, so the dubbed version retains the original feel.</li> <li> <strong>Cost optimization</strong> — Running Whisper + GPT-4o + TTS + Demucs adds up. I optimized by running Demucs locally on Apple Silicon (MPS) instead of cloud GPU.</li> </ul> <h2> What's next </h2> <ul> <li>More voice options and voice cloning</li> <li>Subtitle generation alongside dubbing</li> <li>API for developers</li> <li>Mobile app (PWA)</li> </ul> <p>Would love to hear your feedback. What features would make this useful for you?</p> <p>🔗 <strong><a href="https://tubevoice.io" rel="noopener noreferrer">tubevoice.io</a></strong> — Free tier available (3 free credits to try it out)</p> showdev