Skip to content
Praha LabJoin early access

PRAHA LAB / PUBLIC WORK

Built in
the open.

Praha Cut is our flagship product. Our broader work in speech, visual generation, and agents is documented through public models, datasets, and source code.

Meet Praha Cut ↗
SELECTED WORKLINKED TO THE SOURCE
01
Open model / Voice

PrahaTTS-ML

A public Malayalam text-to-speech LoRA adapter for the Chatterbox non-turbo base model, with its base-model requirements documented.

Model card ↗
02
Open dataset / Speech

Malayalam emotion-balanced speech

A public audio and text dataset for Malayalam speech work with language, speaker gender, and style fields.

Dataset ↗
03
Open model / Visual reasoning

Qwen3.5-4B TikZ LoRA

A published adapter for instruction-to-TikZ generation. Its model card covers training provenance, evaluation, and limitations.

Model card ↗
04
Founder open source / Agents

AutoScribe-CrewAI

Pranav Harshan’s public multi-agent research and writing project with planner, writer, and editor roles.

Source code ↗

More from Praha Lab

Hugging Face ↗GitHub ↗RimaTTS preview ↗