Simpaudio

What does Simpaudio do?

Simpaudio is described in its README as "a free, open-source, offline text-to-speech (TTS) voice studio for Windows. It turns text into natural, studio-quality speech in 8 languages with 40+ realistic voices, powered entirely by Piper TTS and Kokoro TTS running locally on your PC."

The app's feature list goes beyond plain TTS: voice blending (mix two voices to produce a new one), audiobook generation from PDF, EPUB and TXT files, Whisper speech-to-text, a visual SSML editor for pauses, emphasis and pitch, and SRT subtitle export. The README's own one-liner describes it as "the private ElevenLabs alternative that keeps your audio on your machine."

Key facts

• Licence: MIT on the code.
• Platform: Windows 10 or 11 (64-bit). 4 GB RAM is the floor; 8 GB is the recommended figure. 2 GB free disk space.
• TTS engines: Piper (40+ voices, around 60 MB each) and Kokoro (around 340 MB).
• Speech-to-text: OpenAI Whisper (around 150 MB).
• Languages: 8 languages listed by the README.
• Install options: an .exe installer, a portable ZIP, or a Python developer path with python -m venv and pip install -r requirements.txt.
• First-run model downloads are on-demand and cache locally; after that the app runs offline.
• Native Python API: from engines import create_engine; create_engine("Piper TTS").generate(...).
• Runs on CPU; no GPU required.

What does it replace, and where does it fall short?

Simpaudio replaces ElevenLabs for the text-to-speech loop a creator uses: generate narration, build an audiobook, blend voices. Its own README positions it as "the private ElevenLabs alternative" and says audio and text do not leave the PC.

Where it falls short: Windows only, so a macOS or Linux user has to look at piper, kokoro-fastapi or another entry on the open-source alternatives to ElevenLabs list. The voice choice is Piper and Kokoro, which are good open models but not the same as ElevenLabs' hosted voices. The README's "voice cloning" claim is actually "voice blending" (mixing two existing voices); cloning a new speaker from a reference clip is not described. ElevenLabs' API, large hosted catalogue and real-time conversational audio are not here. Marketing claims in the README (voice counts, savings) are the project's own.

How does Simpaudio run?

Download simpaudio-installer.exe and run the setup wizard, or download simpaudio-portable.zip, unzip anywhere (USB, custom folder) and run simpaudio.exe. Developers can clone, create a virtual environment, install requirements and run python app.py. On first use, each model downloads once (Piper voices around 60 MB each, Kokoro around 340 MB, Whisper around 150 MB) and is then cached locally.

The native Python API is a single create_engine call: tts = create_engine("Piper TTS"), followed by tts.generate(...) with the text, voice, speed, pitch, output path and an optional status callback. Batch processing is supported for scripts and audiobook chapters.

Who is Simpaudio for?

A Windows creator who wants to generate narration or audiobook audio without a subscription or a per-minute bill, and keep the files on the machine. A developer building a TTS pipeline in Python can use the engines module directly. Someone on macOS or Linux has to look at piper or kokoro-fastapi on the open-source alternatives to ElevenLabs list. Someone who needs a hosted API or ElevenLabs' large voice catalogue should stay with the paid product.

What limits does the README state?

Windows 10 or 11 (64-bit) only. 4 GB RAM is the minimum, with 8 GB recommended. Models download on first use and are cached locally; the first use needs an internet connection. CPU only is enough; a GPU is not required. The README describes voice blending as mixing two voices, not voice cloning from a reference clip.

Questions people ask

Is Simpaudio open source?

Yes. The code is MIT, an OSI approved licence. The README states this plainly: "The MIT license allows personal, commercial and enterprise use, including in products you sell. The copyright notice must be kept." Contributions go through issues and pull requests on GitHub.

Can Simpaudio be used commercially?

Yes. MIT allows commercial use, modification and redistribution, with the copyright notice travelling with the code. Simpaudio itself is free. The underlying models (Piper, Kokoro, Whisper) each have their own licences on their weights, which the user has to check for the voices they want to ship in a product.

Does Simpaudio work offline?

Yes, after the one-time model download on first use. The README states: "Simpaudio has no cloud component. Your text, your voice models and your generated audio never leave your computer." That is the stated difference from hosted ElevenLabs. The first run needs an internet connection to download the Piper voices, Kokoro and Whisper.

How does Simpaudio compare with ElevenLabs?

ElevenLabs is a hosted TTS platform with a large voice catalogue, voice cloning from a reference clip, an API and a web studio. Simpaudio runs on a Windows machine with Piper and Kokoro voices, with voice blending (mix two existing voices) rather than cloning from a reference. It is free. The open-source alternatives to ElevenLabs page sets it beside Piper, Kokoro, Fish Speech, F5-TTS and Chatterbox.

What hardware does Simpaudio need?

The README names Windows 10 or 11 (64-bit), 4 GB RAM as the floor and 8 GB as the recommended level, with 2 GB of free disk space. The engine runs on CPU; a GPU is not required. First-use model downloads add up to roughly 550 MB for Kokoro, Whisper and one Piper voice; each extra Piper voice is around 60 MB.

Sources

• Simpaudio README and LICENSE: github.com/alimaandev/simpaudio, read 2026-10-26. The README is the source for every engine, language count, model size, install path and limit above.