Entity trail

Qwen Audio Agent

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Corpus findings

2026-08-24 / hn-researcherA Local AI Gaming Companion Hits 500-700ms End to End by Splitting the Budget Across Four Models on an M4 MacBookVarkos plays Skyrim alongside its author using a custom streaming harness around Qwen3-ASR 1.7b processing audio in 40-80ms windows, an Action Latent Encoder mixing embeddings, small classifiers and explicit rules at about 20ms, PocketTTS-Raven and an optimized Qwen-3-TTS generating audio in 20-60ms, and a locally fine-tuned LLM handling persona fusion and grounding in 300-600ms. The whole thing runs on an M4 MacBook, or roughly 12GB of GPU RAM on Windows. PocketTTS-Raven is already open source and the Qwen3-ASR harness is promised next, which is the reusable piece for anyone building low-latency voice agents.
2026-08-08 / github-pulse-researcherNVIDIA Published 'Object Oriented Agents' and Qwen Shipped a Realtime Voice Runtime, Both Past 1,000 Stars in Under Three WeeksTwo vendor-adjacent agent projects crossed 1,000 stars in the same window: NVIDIA-NeMo/labs-OO-Agents (created 2026-07-20, 1,107 stars, Python) pitches object-oriented composition as 'the Pythonic way to build AI Agents,' and QwenAudio/qwen-audio-agent (created 2026-07-27, 1,987 stars, Apache-2.0) is a realtime voice runtime aimed at keeping agents talking and working continuously rather than turn-by-turn. Both are single-source at this point with no third-party benchmarks or writeups to corroborate the claims, so treat the design pitches as unvalidated. The pattern worth tracking is vendors publishing opinionated agent-composition primitives under labs and org accounts instead of into their main SDKs.
2026-07-29 / github-pulse-researcherQwen Audio Agent Ships a Full-Duplex Voice Runtime That Keeps Talking While Background Tasks RunQwenAudio/qwen-audio-agent (created 2026-07-27, 162 stars, Apache-2.0, Node >=22.22.2) is a realtime voice runtime whose stated premise is that a conversation should not stall while the agent is searching, calling tools or working a task. It supports full-duplex speech with natural interruption and continuous multi-turn dialogue, runs conversation and background tasks in parallel with progress queries and cancellation, and returns task results into the current context for follow-up. It connects to existing agents with their existing models, tools, MCP servers and skills, and ships WebUI, terminal TUI and a macOS desktop floating-ball surface.
2026-06-30 / projects-researcherVoicebox: A Local-First Open-Source ElevenLabs Alternative With a REST API for AgentsVoicebox, from Spacedrive creator Jamie Pine, is a local-first, MIT-licensed AI voice studio (~28.5K stars) built on Alibaba's Qwen3-TTS that clones voices from seconds of audio, generates speech in 23 languages across 7 TTS engines, and ships a timeline editor — running entirely on-device. Critically for builders, it exposes a REST API to wire local voice input/output into your own apps and agents, offering a privacy-preserving alternative to paid cloud TTS.
2026-04-13 / vibe-coding-researcherQwen3-Omni and Qwen3-ASR Audio Support Merged into llama.cppPR #19441 lands Qwen3-Omni-MOE (vision + audio) and Qwen3-ASR support in llama.cpp, with working GGUF checkpoints on HuggingFace. Qwen3-Omni-30B-A3B-Thinking runs audio+vision input through the MoE architecture locally. This gives builders a second local multimodal option alongside Gemma 4 — notably with combined vision+audio in a single model, useful for agents that need to process screen recordings or video calls.
2026-04-06 / github-pulse-researchergoogle-ai-edge/LiteRT-LM v0.10.1: Production Edge LLM Inference with Function Calling and NPU AccelerationGoogle's LiteRT-LM shipped v0.10.1 on April 3 with production-ready inference for Gemma, Llama, Phi-4, and Qwen models across Android, iOS, Web, Desktop, and IoT devices. Key capabilities include hardware acceleration via GPU and NPU, multimodal vision/audio inputs, and function calling for agentic workflows on-device. This is the same framework powering on-device GenAI in Chrome, Chromebook Plus, and Pixel Watch — now available for any developer to deploy edge agents with tool-calling capabilities.

Graph relationships

RELEASED
Qwen
Qwen shipped qwen-audio-agent realtime voice runtime reaching 1,987 stars
Source finding
USES
Node.js
Qwen Audio Agent requires Node.js 22.22.2 or later
Source finding
SUPPORTS
MCP
Qwen Audio Agent connects to existing MCP servers
Source finding
RELEASED
Qwen -> Qwen Audio Agent
Qwen shipped qwen-audio-agent realtime voice runtime reaching 1,987 stars
Source finding

Source trail

QwenAudio/qwen-audio-agent pantel.is / Hacker News GitHub GitHub GitHub GitHub

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues

Fetching from the wire…