I Cancelled Wispr Flow & Built This Instead (Free Tool)

I Cancelled Wispr Flow & Built This Instead (Free Tool)

by Pat Simmons

Applied Sciences & Engineering AI — Models & Research UYQArtificial intelligence (AI)UYQFGenerative artificial intelligence
🎙 Pat Simmons 👥 24K 📅 August 20, 2026 ⏱ 15 min 👁 50K 📄 tutorial 🧭 2026-09-07
Available in: English (current) Français

Keywords

Claude Codevoice-to-textlocal modelWispr Flowopen source

Summary

In this tutorial, Pat Simmons demonstrates how to build a free, local alternative to Wispr Flow, a popular voice dictation app. Using Claude Code, he creates a push-to-talk dictation app called Murmur, which runs entirely on the user’s machine, ensuring data privacy. The video walks through the process step-by-step, from the initial prompt to the final polished interface. Key components include a Swift-based HUD, a hotkey system, and a speech-to-text engine. Simmons compares two local speech recognition options: Apple’s native Speech Transcriber and Nvidia’s Parakeet model, downloaded from Hugging Face. He also benchmarks his clone against Wispr Flow, finding that his local solution is faster and equally accurate. The video concludes with a demonstration of the app’s interface, including a transcription history and a customizable dictionary. The creator provides the full code and prompts in a GitHub repository and a blog post, making it easy for viewers to replicate the project.

151 words

Critical Evaluation

Value of the Information & Strength of the Argument

The video provides high practical value by offering a step-by-step guide to building a functional voice dictation app with minimal coding knowledge. The argumentation is solid, supported by live demonstrations and a head-to-head comparison with Wispr Flow. The creator transparently discusses limitations, such as the AX silent failure in Electron apps, and offers solutions. The claim that the clone is faster than Wispr Flow is based on a single test, but the margin is significant, and the methodology is explained. The video also highlights the benefits of local processing, including data privacy and cost savings, which are compelling arguments for the approach.

Scientific Rigor, Source Quality, Title Accuracy

The video demonstrates a rigorous approach by using official APIs and open-source models, and by providing the full code in a public repository. The sources cited include the GitHub repo and the creator’s blog post, which contain detailed instructions and prompts. The title accurately reflects the content, and the video delivers on its promise. The creator acknowledges the strengths of Wispr Flow, such as its marketing and ease of use, which adds balance. The comparison methodology is not fully controlled, but the results are presented with appropriate caveats. Overall, the content is well-sourced and the title is appropriate.

215 words

Title / Content Match

The title accurately reflects the content: the creator cancels Wispr Flow and builds a free alternative, as shown in the video.

Quality & Reliability

7/10

The video is a practical tutorial demonstrating the creation of a local dictation app using Claude Code. The methodology is transparent, with live demonstrations and comparisons. However, the claims about speed and accuracy are based on a single informal test, and the video has a promotional tone for the creator's newsletter and blog.

Chapters

Cited Sources

Concurring Sources

Dissenting Sources

  • Wispr Flow official website — The video claims that Wispr Flow is slower than the local clone, but this is based on a single informal test and may not reflect real-world performance.

Contribution & Novelties

The video offers a novel approach to building a voice dictation app by leveraging Claude Code’s natural language capabilities, making it accessible to non-programmers. It demonstrates the potential of AI coding assistants to rapidly prototype and deploy functional software. The comparison between Apple’s native speech transcriber and Nvidia’s Parakeet provides valuable insights into local speech recognition options.

Pour aller plus loin :

  • Claude Code documentation — Official documentation for Claude Code, the tool used in the video.
  • NVIDIA Parakeet — The Parakeet model page on Hugging Face, as referenced in the video.
  • Apple Speech Framework — Apple’s speech recognition framework, which is the basis for the native transcriber used in the video.

112 words

Radar Profile

The radar profile shows high scores in information quantity and quality, reflecting the detailed tutorial and practical demonstrations. The technical level is moderate, suitable for beginners, while the reliability is good due to the transparent methodology and provided resources.

Reliability 7/10

💬 Positif. Sur les 30 commentaires analysés, la majorité exprime de l'enthousiasme et de la gratitude pour le tutoriel, certains partageant leurs propres expériences de construction d'outils similaires, avec quelques suggestions d'alternatives.