Speech-to-Speech vs meetily
Two of the top projects in Voice, side by side: who leads on each score part and who each one suits, then setup, license, activity and what each review found.
Speech-to-Speech
Modular voice-agent pipeline exposed through the OpenAI Realtime API
meetily
Desktop app that records, transcribes and summarizes meetings locally
Verdict
Speech-to-Speech ranks higher, 66 to 64. Of the 5 score parts we measure for both, Speech-to-Speech leads on 1, meetily on 2 and 2 are even.
- Adoptionmeetily leads by 20 points, 55 to 35: a wide gap.
- FreshnessEven at 100.
- MaintenanceSpeech-to-Speech leads by 30 points, 94 to 64: a wide gap.
- Easy to runEven at 50.
- Agent-readymeetily leads by 10 points, 40 to 30.
Pick Speech-to-Speech if you are
Developers building self-hosted voice agents or Realtime API backends
- Implements core OpenAI Realtime events over WebSocket and WebRTC, so client swaps are easy
- Fully local on Apple Silicon (MLX) or NVIDIA CUDA, with no API key needed
Pick meetily if you are
Professionals who want private, local meeting transcription and notes
- Transcription runs on-device with Whisper or Parakeet; no cloud needed
- Summary provider is selectable, including local Ollama or a custom OpenAI-compatible endpoint
| What we compare | Speech-to-Speech | meetily |
|---|---|---|
| Score parts, out of 100 | ||
| Adoption | 35, known | 55, popular |
| Freshness | 100, active | 100, active |
| Maintenance | 94, healthy | 64, fair |
| Easy to run | 50, easy | 50, easy |
| Agent-ready | 30, minimal | 40, minimal |
| Facts from GitHub and the README | ||
| Stars | 13.5k | 31.7k |
| License | Apache-2.0 (permissive) | MIT (permissive) |
| Last commit | Oct 2026 | Sep 2026 |
| Last release | Sep 2026 | Sep 2026 |
| Language | Python | Rust |
| Docker | Yes | No |
| GPU | Optional | Optional |
| arm64 or Apple Silicon | Mentioned | Mentioned |
Speech-to-Speech
Runs a VAD, STT, LLM and TTS cascade, with each stage in its own thread and every backend swappable by CLI flag. It serves the core OpenAI Realtime event set over WebSocket and WebRTC, so existing Realtime clients can point at it. Defaults are Parakeet TDT for speech recognition and Qwen3-TTS for speech output, with the LLM running locally or through any OpenAI-compatible endpoint.
Who it is for: Developers building self-hosted voice agents or Realtime API backends
Strengths
- Implements core OpenAI Realtime events over WebSocket and WebRTC, so client swaps are easy
- Fully local on Apple Silicon (MLX) or NVIDIA CUDA, with no API key needed
- Many interchangeable STT and TTS backends, including Whisper, Kokoro, Pocket TTS and OmniVoice
- Apache-2.0, installable from PyPI, with a packaged microphone client
Weaknesses
- Qwen3-TTS GGML wheel targets CUDA 12.8 and glibc 2.39 by default
- Fully local NVIDIA setup budgets about 24 GB VRAM; the README calls this an estimate
- Only the core Realtime event set is implemented, not the full API
- Some extras conflict, e.g. DeepFilterNet needs numpy<2 while Pocket TTS needs numpy>=2
- RAM ≥ 16 GB
- GPU optional
- Docker + Compose
- Needs OpenAI-compatible LLM server (optional), PortAudio and libsndfile on Ubuntu
- Models: Parakeet TDT, Qwen3-TTS, Whisper, Kokoro-82M, Transformers LLMs
- port 8765
meetily
Meetily is a Tauri desktop app with a Rust backend and a Next.js frontend. It captures microphone and system audio, transcribes on-device with Whisper or Parakeet models, and generates summaries through Ollama, Claude, Groq, OpenRouter, or any OpenAI-compatible endpoint. Installers exist for Windows and macOS; Linux requires a source build.
Who it is for: Professionals who want private, local meeting transcription and notes
Strengths
- Transcription runs on-device with Whisper or Parakeet; no cloud needed
- Summary provider is selectable, including local Ollama or a custom OpenAI-compatible endpoint
- Records microphone and system audio together with mixing
- Imports existing audio files and re-transcribes with another model or language (beta)
Weaknesses
- No prebuilt Linux package; must build from source
- Speaker diarization is not in the community edition; planned for PRO
- Custom summary templates, advanced exports and auto-join are PRO-only
- Standard Windows installer uses Vulkan; CUDA needs a source build
- GPU optional
- Needs Ollama (optional, for local summaries)
- Models: Whisper, Parakeet (ONNX), Ollama, Claude, Groq
More in Voice
Building with these?
We help teams pick an open-source AI stack and run it in production. Client work never changes a rank.