FAQ

General

Q: Can I use Vocoding without internet?

Yes, partially. Whisper transcription is 100% local. If you use Ollama in local mode, you don't need internet either. If you use Ollama cloud, Groq, or OpenRouter, you will need an internet connection.

Q: Do I need a powerful GPU?

On Apple Silicon Macs, the integrated GPU is used automatically — no dedicated GPU needed. On Intel Macs and Windows, Whisper runs on CPU. A GPU is not required.

Q: How much disk space do I need?

The app itself needs ~500 MB. Each Whisper model requires additional space (75 MB to 2.9 GB depending on the model).

---

macOS Specific

Q: I have an old MacBook Pro from 2015, will it work?

Yes, as long as it runs macOS 10.15 (Catalina) or newer. It will use CPU for transcription, so we recommend lighter Whisper models (tiny, base, or small).

Q: I have an M4 Mac, is it compatible?

Yes, all Apple Silicon chips (M1, M2, M3, M4 and their variants) are supported with full GPU acceleration.

Q: Do I need to grant accessibility permissions?

Microphone permission is required. Accessibility permission is optional but needed for the auto-paste feature to work.

---

Windows Specific

Q: When will the Windows version be available?

The Windows version is in active development. Check vocoding.com for updates on availability.

Q: Can I already use Ollama on Windows while waiting?

Yes! Ollama is independently available for Windows. You can start using Ollama now and it will work with Vocoding when the Windows version is ready.

Q: Do I need Windows 11?

Windows 10 (x64) is supported, but Windows 11 is recommended because WebView2 Runtime comes pre-installed.

---

Performance

Q: Which Whisper model should I use?

It depends on your hardware:

• Apple Silicon: large-v3-turbo for best balance of speed and accuracy
• Intel Mac: small or medium for good balance on CPU
• Windows (when available): small or base for CPU-based transcription
• Quick notes on any platform: tiny or base

Q: How much RAM does Ollama need?

It depends on how you use it:

• Ollama Cloud: No extra RAM needed — processing happens on Ollama's servers.
• Ollama Local: Depends on the model size:
  • 8B models: ~4-5 GB RAM
  • 70B models: ~40 GB RAM
  • Smaller quantized models need less

Q: What's the difference between Ollama local and cloud?

Ollama Local runs models directly on your machine (free, offline, but limited by your hardware). Ollama Cloud runs state-of-the-art models on their servers with three plans: Free ($0, basic cloud access with limited usage), Pro ($20/month, multiple simultaneous cloud models and 3 private models), or Max ($100/month, 5+ simultaneous models, 5x Pro usage, and 5 private models). Vocoding supports all 237+ cloud-available models. Most users prefer cloud mode because running cutting-edge models locally requires extremely expensive hardware.