Yougu Speech · Realtime · WebSocket

English Realtime Evaluation · Streaming

WebSocket streaming: read aloud and get a live transcript, follow-along score, coverage and rate; stopping yields the full six-dimension result. ~20ms PCM frames, millisecond feedback. Use the mic live, or upload audio for file-simulated streaming.

readstreamlivetranscriptscore/coverageratewpmWSstreaming

Configuration 评测配置

Reference text (follow-along)

Examples 示例

未选择音频

Streaming · Live transcript

Click “● Start” on the left — your words, a live follow-along score and coverage stream here in real time…

The live score is for follow-along only; stopping produces the full six-dimension final evaluation (per-word annotation, sentence scores, AI report).

Generating · 全量评测生成中

正在为你生成评测报告

识别朗读内容

逐字声学评分

AI 多模态点评

正在听你的录音…

评测结果 Result

记录号 Record ID

—

句尾语调 Rear Tone

—

逐字标注 Per-character

优 ≥85良 ≥70中 ≥55差 <55漏读增读 / 重复

对比播放 Compare

考生录音

标准示范音语速

How streaming works 实时流式

01 Connect

Open WS

A warm WebSocket to the engine avoids cold-start latency.

02 Stream

Stream PCM

Mic audio downsampled to 16k, ~20ms frames (640B PCM16).

03 Live

Live feedback

Word-by-word transcript + live follow-along score + coverage + rate.

04 Final

Stop = final

Stop sends stop; full six-dimension result + annotation + AI report.