Nicheloom

Market intelligence for builders — see what's gaining traction before it's crowded.

EarShot

实时面试提词器:本地 ASR + BM25 检索 + LLM 快答,说完 1.3~1.5 秒出答案(Windows)

Details

External ID
1367483350
Source
GITHUB
Company
—
Product
EarShot
Website domain
github.com
Launched
Sept. 12, 2026
Cohort
—
Upvotes
10
Upvotes percentile
0.28183448629259544
Tags
asr, bm25, interview, onnx, pyqt6, realtime, sensevoice, teleprompter, windows
Fetched at
Sept. 16, 2026, 5:02 p.m.
Updated at
Sept. 16, 2026, 5:02 p.m.

Enrichment

Theme
niche developer utilities and guides
Vertical
Horizontal
Function
Agent / copilot
Audience
Prosumer
AI stance
AI-native
Project type
Hobby / open-source project
Normalized one-liner
real-time interview copilot and teleprompter
Manually corrected
False

Could you build this?

Partial While an LLM can script the integration of ASR, BM25, and LLM APIs, achieving sub-1.5s real-time end-to-end latency on Windows requires strict pipeline optimization and audio capture tuning.

What it would actually take: The stack likely uses Python or C++ with system audio capture (WASAPI/virtual loopback), lightweight local streaming ASR (e.g., Sherpa-ONNX or optimized Whisper), in-memory BM25 retrieval, and low-latency streaming LLM inference (e.g., vLLM or Groq). The hard part is minimizing the audio buffer chunk latency, silence/endpoint detection, and concurrent async pipelining to hit consistent 1.3s response times without hallucinating or dropping audio frames.

Competitors

Other products that read as similar to this one — 342 launches clear the similarity bar, closest 8 shown.

Attention rank: #243 of 343 (itself plus its competitors, highest first — normalized so YC and Product Hunt are compared fairly).

Launched 315 days after the earliest competitor.

Other launches for this product

Same idea, different domain

Nobody's really built a agent / copilot tool for Agriculture yet.