Speech AI: 14 solutions compared
Which Speech AI tools can you self-host, run on a Mac or connect to? FrontierStack's catalogue covers 14 Speech AI options: CLOVA Speech (NAVER Cloud), CLOVA Voice / CLOVA Dubbing TTS (NAVER Cloud), Coqui TTS, Faster Whisper, Handy, Kokoro TTS and others. For 14 of them FrontierStack installs the service on a Mac or server, or connects to one you already run, and checks status, ports and certificates where they apply.
| Name | Type | What FrontierStack does |
|---|---|---|
| CLOVA Speech (NAVER Cloud) | Self-hosted or connected | Install or connect; status, ports and certificate checks where they apply |
| CLOVA Voice / CLOVA Dubbing TTS (NAVER Cloud) | Self-hosted or connected | Install or connect; status, ports and certificate checks where they apply |
| Coqui TTS | Self-hosted or connected | Install or connect; status, ports and certificate checks where they apply |
| Faster Whisper | Self-hosted or connected | Install or connect; status, ports and certificate checks where they apply |
| Handy | Self-hosted or connected | Install or connect; status, ports and certificate checks where they apply |
| Kokoro TTS | Self-hosted or connected | Install or connect; status, ports and certificate checks where they apply |
| Meetily | Self-hosted or connected | Install or connect; status, ports and certificate checks where they apply |
| Piper TTS | Self-hosted or connected | Install or connect; status, ports and certificate checks where they apply |
| Scriberr | Self-hosted or connected | Install or connect; status, ports and certificate checks where they apply |
| VibeVoice | Self-hosted or connected | Install or connect; status, ports and certificate checks where they apply |
| Voicebox | Self-hosted or connected | Install or connect; status, ports and certificate checks where they apply |
| Whisper | Self-hosted or connected | Install or connect; status, ports and certificate checks where they apply |
| Whisper.cpp | Self-hosted or connected | Install or connect; status, ports and certificate checks where they apply |
| Wispr Flow | Self-hosted or connected | Install or connect; status, ports and certificate checks where they apply |
CLOVA Speech (NAVER Cloud)
NAVER CLOVA long-form speech recognition
Registering a CLOVA Speech app yields an Invoke URL (https://clovaspeech-gw.ncloud.com/external/v1/{service-id}/{app-id}) and a Secret Key sent as X-CLOVASPEECH-API-KEY. The short-sentence STT (CSR) instead uses API Gateway Client ID/Secret.
Install or connect CLOVA Speech (NAVER Cloud); then check its status, ports and certificate from FrontierStack. The same screen links to firewall checks and backups where they apply.
CLOVA Voice / CLOVA Dubbing TTS (NAVER Cloud)
NAVER CLOVA Voice premium Korean/English TTS
Uses the same Application Client ID/Secret style as Maps/Papago. Endpoint https://naveropenapi.apigw.ntruss.com/tts-premium/v1/tts answered 401 for bogus keys (live). Newer per-product hosts may exist.
Install or connect CLOVA Voice / CLOVA Dubbing TTS (NAVER Cloud); then check its status, ports and certificate from FrontierStack. The same screen links to firewall checks and backups where they apply.
Coqui TTS
Open-source text-to-speech & voice cloning
Coqui TTS (🐸TTS) is a deep-learning toolkit for text-to-speech; many pretrained voices/languages, multi-speaker models and voice cloning (XTTS). pip install TTS, then tts --text "hello" --out_path out.wav, or run tts-server for an HTTP API on :5002.
Install or connect Coqui TTS; then check its status, ports and certificate from FrontierStack. The same screen links to firewall checks and backups where they apply.
Faster Whisper
CTranslate2 Whisper; up to 4× faster
faster-whisper reimplements Whisper on CTranslate2 for up to 4× faster transcription at lower memory, with word-level timestamps and batching. pip install faster-whisper; load a model and transcribe from Python. Powers many real-time captioning and subtitle pipelines.
Install or connect Faster Whisper; then check its status, ports and certificate from FrontierStack. The same screen links to firewall checks and backups where they apply.
Handy
Free local push-to-talk speech-to-text (Whisper)
Handy is a free, open-source, cross-platform speech-to-text app; press a shortcut, speak, and it types the transcription into whatever app you're in. It runs entirely on-device using Whisper (whisper.cpp / Parakeet), so audio never leaves your Mac. Install with Homebrew (brew install --cask handy), then grant Microphone and Accessibility permissions, choose a model and set your push-to-talk hotkey in the app. Configure those below.
Install or connect Handy; then check its status, ports and certificate from FrontierStack. The same screen links to firewall checks and backups where they apply.
Kokoro TTS
Small, fast, high-quality open-weight TTS (82M)
Kokoro is a tiny (82M-parameter) open-weight text-to-speech model that punches far above its size; natural, expressive voices at very low latency, runnable on CPU or a modest GPU, even in the browser. Weights are Apache-2.0. Use it from Python (pip install kokoro, plus espeak-ng for phonemes), via the fast kokoro-onnx runtime, or run the OpenAI-compatible Kokoro-FastAPI server for a drop-in /v1/audio/speech endpoint. Great as a local, license-friendly voice for narration and assistants. See the alternative repos below for the ONNX and server options.
Install or connect Kokoro TTS; then check its status, ports and certificate from FrontierStack. The same screen links to firewall checks and backups where they apply.
Meetily
Local AI meeting notetaker (desktop app)
Meetily is a privacy-focused AI meeting assistant that transcribes and summarises meetings entirely on your machine (bundled Whisper). It's a Tauri desktop app for macOS/Windows (Linux from source); not a hosted server, so there's no web UI or port. MIT.
Install or connect Meetily; then check its status, ports and certificate from FrontierStack. The same screen links to firewall checks and backups where they apply.
Piper TTS
Fast, local neural text-to-speech
Piper is a fast, lightweight neural TTS that runs well on a Raspberry Pi or any Mac; high-quality offline voices, the engine behind Home Assistant's local voice. Download a voice model (.onnx) and pipe text: echo 'hello' | piper -m voice.onnx -f out.wav.
Install or connect Piper TTS; then check its status, ports and certificate from FrontierStack. The same screen links to firewall checks and backups where they apply.
Scriberr
Self-hosted audio transcription app (WhisperX) with speaker diarization
Scriberr is a self-hosted transcription web app: upload audio (or record in the browser), and it transcribes locally with the WhisperX engine; fast Whisper inference with word-level timestamps and optional speaker diarization; then lets you summarize transcripts with Ollama or an OpenAI key using your own prompts. Runs via Docker (ghcr.io/rishikanthc/scriberr); the web UI listens on :8080. Everything stays on your hardware; meetings, interviews and voice notes never touch a cloud transcription service unless you choose a cloud summarizer.
Install or connect Scriberr; then check its status, ports and certificate from FrontierStack. The same screen links to firewall checks and backups where they apply.
VibeVoice
Microsoft's long-form, multi-speaker TTS
VibeVoice is Microsoft's open-source frontier text-to-speech model for long-form, multi-speaker audio; expressive, podcast-style conversations with up to four distinct speakers and natural turn-taking. It pairs an LLM with a diffusion head and a low-frame-rate acoustic tokenizer to stay coherent over long durations. Run it locally from the GitHub repo / Hugging Face weights (a 1.5B model and larger variants; a GPU helps): install the package, then synthesize from a script of speaker turns. Good for narration, dialogue and synthetic voice-over. Weights are MIT-licensed and for research use.
Install or connect VibeVoice; then check its status, ports and certificate from FrontierStack. The same screen links to firewall checks and backups where they apply.
Voicebox
Open-source local voice-to-text for macOS
Voicebox (jamiepine/voicebox) is an open-source macOS voice-to-text app; press a hotkey, speak, and it transcribes locally. Grab the latest build from the GitHub releases (or clone and build from source), then grant Microphone & Accessibility permissions. As a notarized .app download rather than a Homebrew package, it has no one-click brew install; use the download button below.
Install or connect Voicebox; then check its status, ports and certificate from FrontierStack. The same screen links to firewall checks and backups where they apply.
Whisper
OpenAI's speech-to-text model (Python)
Whisper is OpenAI's open-source automatic speech-recognition model; reliable multilingual transcription and translation. pip install -U openai-whisper (needs ffmpeg), then whisper audio.mp3 --model small. Runs locally on CPU/GPU; for faster/leaner inference see Whisper.cpp or Faster-Whisper.
Install or connect Whisper; then check its status, ports and certificate from FrontierStack. The same screen links to firewall checks and backups where they apply.
Whisper.cpp
Fast C/C++ Whisper inference (CPU/Metal)
whisper.cpp is a dependency-free C/C++ port of Whisper with Apple Silicon (Metal/CoreML) acceleration; great for on-device transcription with no Python. brew install whisper-cpp; download a GGML model and run whisper-cli -m model.bin audio.wav.
Install or connect Whisper.cpp; then check its status, ports and certificate from FrontierStack. The same screen links to firewall checks and backups where they apply.
Wispr Flow
AI voice dictation that types into any app
Wispr Flow (a.k.a. WhisperFlow) is an AI dictation app for macOS; hold a hotkey, speak, and it types polished, auto-punctuated text into whatever app you're in, with voice commands and editing. Install the desktop app with Homebrew (brew install --cask wispr-flow), then grant Microphone & Accessibility permissions and set your push-to-talk hotkey in the app. It's a cloud-assisted dictation service, so sign in inside the app.
Install or connect Wispr Flow; then check its status, ports and certificate from FrontierStack. The same screen links to firewall checks and backups where they apply.
Run it from your Mac.
FrontierStack installs, monitors and secures services on this Mac and on linked servers; and keeps your passwords and keys out of any AI's sight.
Download FrontierStackApple notarized · Safe & secure · macOS 13 Ventura+
