GUIDE

An alternative to Dia, Orpheus and Sesame CSM that runs on a CPU

Dia, Orpheus TTS and Sesame CSM brought large language model ideas to speech: lively dialogue, laughter, emotion tags and fast streaming. They also share a requirement: an NVIDIA graphics card, and for Dia, English text only.

The free SuperClonica engine runs the smaller Supertonic 3 model on an ordinary processor, in 31 languages, with a catalog of ready voices. This guide compares them honestly and says when one of the three is the better pick.

Browse voices Install the SuperClonica engine

What the three do well

For expressive English conversation on a graphics card, they are among the most impressive open models.

  • Dia, 1.6 billion parameters, writes whole dialogues from a script with [S1] and [S2] speaker tags and can laugh or cough; it supports English only and has been tested on GPUs.
  • Orpheus TTS is built on a 3 billion parameter Llama model, takes emotion tags and streams with about 200 ms latency; it runs on vLLM, on a graphics card.
  • Sesame CSM, 1 billion parameters, needs a CUDA graphics card and access to the Llama 3.2 1B model.
  • All three are under the Apache 2.0 license.

Compare the sound free

On the page of any voice, paste a line you voiced with Dia, Orpheus or CSM, 30 to 100 characters, choose the language and press Voice it. In a few seconds you can compare, with no account.

What changes with SuperClonica

Long texts are split at natural pauses and joined into one file, so a chapter or a script goes in as one call.

  • Install it with one command: python -m pip install superclonica-cpu, or run the Docker image as an HTTP service.
  • It runs on an ordinary processor: in our tests a short phrase took 0.43 s on four threads of a Ryzen 7 5800X.
  • 31 languages, including Spanish, German, French, Russian, Japanese and Korean; Chinese is not supported yet.
  • The text is read as written, and the same seed gives the same take again.

Voices: prompts, presets and trained voices

Dia keeps a speaker by an audio prompt or a fixed seed; Orpheus comes with preset voices; CSM takes context audio. SuperClonica has ten stock voices and more than 170 catalog voices, each with a free demo; for your own voice, train it once in the SuperClonica studio from a clean 5 to 10 second recording of a speaker who agreed, on a computer with an NVIDIA graphics card, or order it on the exchange. After that the voice runs on any processor and in all 31 languages.

Which one to choose

Choose Dia, Orpheus or CSM for expressive English conversation with laughter and emotion tags, or for a voice agent with streaming, when you have a graphics card for every request. Choose SuperClonica for clean speech in 31 languages on ordinary servers and laptops, for ready character and narrator voices, and for large volumes without graphics cards.