Story

On air. The free wire runs 15 minutes behind live. See pricing

World, crypto, finance and tech news, scored for impact.

TechHacker NewsFiled 4 h ago

Rebuilding my voice assistant pipeline

Open the sourceReported by Hacker News, blog.platypush.tech, 19 November 2026

Not investment advice. Muhabbit scores how much a story may matter, not which way to trade.

Platypush creator Fabio Manganiello has rebuilt his self-hosted voice assistant pipeline, swapping the previous setup for a new stack built around the local assistant.microwakeword plugin for wake-word detection, OpenAI's speech-to-text and text-to-speech plugins for cloud processing, and the openai plugin for response generation, with local fallbacks including assistant.vosk and tts.piper. The change follows roughly two months of using the prior pipeline as his main driver, during which he reported high false-positive and false-negative rates and poor detection of speech start and end, particularly in noisy environments and with cheap microphones. The revised configuration pairs local wake-word detection with cloud-based recognition and synthesis, while retaining local options for speech-to-text and text-to-speech. The rebuild matters because it highlights the trade-offs between commercial-grade robustness and self-hosted privacy and control in voice assistants. Next, watch whether the fine-tuned pipeline holds up in sustained daily use and whether the local fallbacks see broader adoption.

Muhabbit's AI wrote this summary in its own words from Hacker News's report. It can get details wrong, so check the source.

Named in this story

Platypush, OpenAI, Picovoice, Piper, Vosk