DEDUPED Hot AI repos. Real use cases. What people build. Get the Monday 5

BEST AI REPOS BY TASK

Best open-source AI repos for voice, speech-to-text and text-to-speech (2026)

7 repos compared · updated 2026-10-27

What is the best open-source repo for AI voice, transcription and text-to-speech?

The best open-source AI repos for voice, speech-to-text and text-to-speech right now are Openwhispr, Kitten TTS and LocalAI. For dictation on your own computer use OpenWhispr; for small, fast text-to-speech models use Kitten-TTS; for one local server that also handles voice next to LLMs use LocalAI.

Comparison

RepoWhat it doesStarsStart here
Openwhispr
OpenWhispr/openwhispr · JavaScript		OpenWhispr is a free, open-source alternative to WisprFlow and Granola.10,055Dictate into any app with private, offline speech-to-text Beginner, 10 minutes
Kitten TTS
KittenML/KittenTTS · Python		Kitten-TTS is an open-source family of text-to-speech models, including the 1.7B 1-bit Kitten-TTS 2 with voice cloning and expression control and lightweight 15M to 80M models that run on CPU.16,217Add natural local speech to a Python app in a few lines Beginner, 15 minutes
LocalAI
mudler/LocalAI · Go		LocalAI is an open-source AI engine that runs LLMs, vision, voice, image and video models on any hardware, with no GPU required.49,983Self-host an OpenAI-compatible API on your own server Beginner, 30 minutes
PersonaPlex
NVIDIA/personaplex · Python		PersonaPlex is NVIDIA's real-time, full-duplex speech-to-speech conversational model based on Moshi that controls persona through text role prompts and audio voice conditioning.11,246Talk to a full duplex voice agent in your browser Advanced, 45 minutes
Speech CPP
nyosegawa/speech.cpp · C++		speech.cpp is a C++ command and library on ggml for speech synthesis, speech recognition and voice activity detection, running on the CPU, Metal or Vulkan with an OpenAI-compatible server.195Transcribe long Japanese recordings with timestamps Intermediate, 30 minutes
Voice Studio
debpalash/VoiceStudio · Python		VoiceStudio is an open-source, local desktop app and API for voice cloning, voice design, video dubbing, dictation, transcription and audiobook creation.81,165Dub a video into timed speech locally Beginner, 45 minutes
Meetcap
seacen/meetcap · Shell		MeetCap is a Mac meeting assistant that records and transcribes meetings on your computer, writes live summaries and minutes, and lets an AI agent start recordings and search past meetings.142Find what was agreed in past meetings by asking your agent Intermediate, 1 hour

When to pick which

Openwhispr OpenWhispr/openwhispr

OpenWhispr is a free, open-source alternative to WisprFlow and Granola. Press a hotkey, speak and your words appear at your cursor, using fully offline speech-to-text models such as Whisper and NVIDIA Parakeet or cloud providers. It also transcribes meetings with speaker diarization, keeps searchable notes, offers a Voice Assistant hotkey that sends commands to GPT, Claude, Gemini or local models, and has a public API and MCP server.

Best for: Writers, developers and professionals on macOS, Windows or Linux who want private voice dictation and meeting notes without a subscription.

Needs: macOS, Windows or Linux; Optional: local models download for offline use, or cloud provider keys; To run from source: Node.js 24+

Start here: Dictate into any app with private, offline speech-to-text Beginner, 10 minutes

  1. Download the app for your platform, or on macOS run brew install --cask openwhispr
  2. Choose a local speech-to-text model such as Whisper or NVIDIA Parakeet
  3. Set your global dictation hotkey
  4. Press the hotkey, speak, and the text appears at your cursor

All 3 use casesGitHub

Kitten TTS KittenML/KittenTTS

Kitten-TTS is an open-source family of text-to-speech models, including the 1.7B 1-bit Kitten-TTS 2 with voice cloning and expression control and lightweight 15M to 80M models that run on CPU.

Best for: Developers adding speech to apps, assistants or content pipelines who want local, CPU-friendly TTS.

Needs: Python with pip; A CPU is enough for the lightweight models; Kitten-TTS 2 runs realtime on CPUs and consumer GPUs per the README

Start here: Add natural local speech to a Python app in a few lines Beginner, 15 minutes, works with claude-code, codex, cursor

  1. Install: pip install kittenml
  2. Load the model: m = KittenTTS("KittenML/kitten-tts-2")
  3. List voices with m.available_voices and pick one
  4. Save speech: m.generate_to_file("Hello, world.", "output.wav", voice="Bruno")

All 3 use casesGitHub

LocalAI mudler/LocalAI

LocalAI is an open-source AI engine that runs LLMs, vision, voice, image and video models on any hardware, with no GPU required. Each backend (llama.cpp, vLLM, whisper.cpp, stable-diffusion, MLX and more) is pulled only when a model needs it, and every backend sits behind drop-in OpenAI, Anthropic and ElevenLabs compatible APIs, with multi-user auth, quotas and built-in agents with tools, RAG and MCP.

Best for: Developers and teams who want a private, self-hosted replacement for cloud AI APIs that existing OpenAI or Anthropic clients can talk to.

Needs: Docker or Podman, or the macOS app (not signed by Apple); Optional GPU: NVIDIA, AMD, Intel, Apple Silicon or Vulkan; CPU-only works; Disk space for the models you pull

Start here: Self-host an OpenAI-compatible API on your own server Beginner, 30 minutes, works with claude-code, codex, cursor

  1. Start the CPU image: docker run -ti --name local-ai -p 8080:8080 localai/localai:latest
  2. Or with an NVIDIA GPU: docker run -ti --name local-ai -p 8080:8080 --gpus all localai/localai:latest-gpu-nvidia-cuda-12
  3. Load a model, for example local-ai run llama-3.2-1b-instruct:q4_k_m
  4. Point your app's OpenAI base URL at http://localhost:8080

All 3 use casesGitHub

PersonaPlex NVIDIA/personaplex

PersonaPlex is NVIDIA's real-time, full-duplex speech-to-speech conversational model based on Moshi that controls persona through text role prompts and audio voice conditioning.

Best for: Speech AI researchers and developers building low-latency voice agents with a consistent persona.

Needs: Linux with an NVIDIA GPU; Opus development library; A Hugging Face token with access to the model

Start here: Talk to a full duplex voice agent in your browser Advanced, 45 minutes

  1. Install the Opus codec: sudo apt install libopus-dev on Ubuntu or Debian
  2. From the repo run pip install moshi/.; Blackwell GPUs also need the CUDA 13 PyTorch wheels from the README
  3. Accept the PersonaPlex model license on Hugging Face and export your HF_TOKEN
  4. Start the server: SSL_DIR=$(mktemp -d); python -m moshi.server --ssl "$SSL_DIR"

All 3 use casesGitHub

Speech CPP nyosegawa/speech.cpp

speech.cpp is a C++ command and library on ggml for speech synthesis, speech recognition and voice activity detection, running on the CPU, Metal or Vulkan with an OpenAI-compatible server.

Best for: Developers and power users who want local speech synthesis and transcription without Python or CUDA, and who need text-to-speech, transcription or an audio API on their own machine.

Needs: macOS on Apple silicon, Linux on x86-64, or Windows x64; Internet access to fetch the installer and models from Hugging Face the first time; Free disk space for the model files, for example 1.21 GB for qwen3-tts-0.6b; A terminal (PowerShell on Windows)

Start here: Transcribe long Japanese recordings with timestamps Intermediate, 30 minutes

  1. Install speech with curl -fsSL https://raw.githubusercontent.com/nyosegawa/speech.cpp/main/install.sh | sh on macOS or Linux, or irm https://raw.githubusercontent.com/nyosegawa/speech.cpp/main/install.ps1 | iex in PowerShell on Windows.
  2. Put the recording in a WAVE file, for example meeting.wav.
  3. Run speech asr reazonspeech-v2 --timestamps meeting.wav. The model is fetched the first time.
  4. Check the first few lines of the output against the audio, since the model is for Japanese only.

All 5 use casesGitHub

Voice Studio debpalash/VoiceStudio

VoiceStudio is an open-source, local desktop app and API for voice cloning, voice design, video dubbing, dictation, transcription and audiobook creation.

Best for: Content creators, localization staff and developers who want speech and dubbing tools that run on their own hardware.

Needs: A computer with an NVIDIA GPU, Apple Silicon, or a CPU (slower); Free disk space; the CPU build of PyTorch needs about 5 GB; Permission to use any voice you clone; Internet access for the installer and the first model download

Start here: Dub a video into timed speech locally Beginner, 45 minutes

  1. curl -fsSL https://voicestudio.sh/install | sh
  2. Open Voice Studio and choose the Video dubbing workspace.
  3. Add your video and the voice or language you want, then install the required model when prompted.
  4. Generate the dub and listen to the sections where speech starts and stops to check timing.

All 6 use casesGitHub

Meetcap seacen/meetcap

MeetCap is a Mac meeting assistant that records and transcribes meetings on your computer, writes live summaries and minutes, and lets an AI agent start recordings and search past meetings.

Best for: People who attend many meetings, need minutes they can trust, or want an AI agent to search and summarize what was said.

Needs: A Mac with macOS 14.2 or later, Apple silicon or Intel; Permissions granted during MeetCap setup (covered in the docs at meetcap.seacen.dev); An AI agent such as Claude Code or Codex for agent-driven use cases, or a local model, your own API key, or a compatible agent for minutes

Start here: Find what was agreed in past meetings by asking your agent Intermediate, 1 hour, works with claude-code, codex

  1. Download MeetCap from https://meetcap.seacen.dev/download, open it and follow the short setup, or run curl -fsSL https://meetcap.seacen.dev/install.sh | bash in the terminal.
  2. Record your meetings with MeetCap from the menu bar, a hotkey, your agent, or a schedule so that transcripts and minutes are saved.
  3. Load the MeetCap agent skill from the seacen/meetcap repository into your agent, such as Claude Code or Codex.
  4. Ask your agent a question about a past decision, for example what price was agreed with a supplier last month.

All 5 use casesGitHub

Use cases for voice, speech-to-text and text-to-speech

Questions

What is the best open-source repo for AI voice, transcription and text-to-speech?

The best open-source AI repos for voice, speech-to-text and text-to-speech right now are Openwhispr, Kitten TTS and LocalAI. For dictation on your own computer use OpenWhispr; for small, fast text-to-speech models use Kitten-TTS; for one local server that also handles voice next to LLMs use LocalAI.

Which repo for voice, speech-to-text and text-to-speech is easiest to start with?

Openwhispr: its starter use case "Dictate into any app with private, offline speech-to-text" is rated Beginner and takes about 10 minutes.

How do I get started with Openwhispr?

Download the app for your platform, or on macOS run brew install --cask openwhispr; Choose a local speech-to-text model such as Whisper or NVIDIA Parakeet; Set your global dictation hotkey. Full steps and a copy-paste agent prompt: https://dedupedai.com/use-cases/openwhispr-openwhispr/dictate-into-any-app-privately/

Are these voice, speech-to-text and text-to-speech repos free?

They are open-source projects on GitHub (check each license; some, like fair-code projects, limit commercial use). Some also sell a hosted version or need a paid model API key; the requirements for each repo are listed on this page.

How is this list of voice, speech-to-text and text-to-speech repos chosen?

Deduped tracks the hottest AI repos on GitHub and writes step-by-step use cases from each README. For this page we picked the repos that do this job, ordered by fit and then GitHub stars. Repos we only inferred from a short README, or that we do not index, are left out. The page is rebuilt with the site; stars are as of the last build in 2026.

More answers

For AI agents: Markdown version · JSON · MCP server · llms.txt

Third-party repos we don't own. AI-written use cases can be wrong. Disclaimer