Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add bleakbelladonnals/asr-transcription-skills --skill transcription-quality-evaluationgit clone --depth 1 https://github.com/bleakbelladonnals/asr-transcription-skillsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/bleakbelladonnals/asr-transcription-skills/transcription-quality-evaluation)<a href="https://agentmods.dev/skills/bleakbelladonnals/asr-transcription-skills/transcription-quality-evaluation"><img src="https://agentmods.dev/badge/skills/bleakbelladonnals/asr-transcription-skills/transcription-quality-evaluation/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/bleakbelladonnals/asr-transcription-skills/transcription-quality-evaluation"><img src="https://agentmods.dev/badge/skills/bleakbelladonnals/asr-transcription-skills/transcription-quality-evaluation.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00154 | $0.02197 |
| Opus 5 | $0.00077 | $0.01099 |
| Sonnet 5 | $0.00031 | $0.00439 |
| Haiku 4.5 | $0.00015 | $0.00220 |
Grade A, and why
transcription-quality-evaluation scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 12d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 81 lines — stays where its author put it; the contents beside it link to each section on GitHub.
录音转写质量对比评测
何时使用
- 用户想比较多个转写方案(录音硬件端到端 / 云转写产品 / API 模型)谁转得准、谁录得好、钱花得值不值
- 用户提供了同一时段的录音文件 + 各家转写 SRT,要求对比音频质量和转写质量
- 输出:对比矩阵(音频质量表 + 转写质量表 + 产品可用性 + 成本 + 结论),可直接改写成评测文章/产品分析作业
核心原则:拆开测,别混着测
"录音硬件+云转写"产品的贵,贵在「采集端 + 转写端 + 总结归档」一条龙。必须拆成两个变量:
- 采集端(硬件麦克风/DSP)→ 用 ffmpeg 客观指标比
- 转写端(引擎/模型)→ 同一段音频过不同引擎比
- 端到端方案(录音豆/录音卡 app 只转自己录的音)无法喂外部音频 → 补一层"同一音频上传纯转写产品"(通义听悟/讯飞听见),剥离硬件差异
- 理想测试矩阵 = 2×2(音频源 × 引擎);免费额度内完成(3
5 段 × 510 分钟即可,故意埋术语/数字"地雷"算保留率)
一、音频质量客观指标(ffmpeg,全部可复现)
# 基本参数:编码/采样率/声道/时长/码率
ffprobe -v quiet -show_format -show_streams in.mp3 | grep -E "duration|bit_rate|sample_rate|channels|codec_name"
# 响度与动态范围(EBU R128):I=整体响度(LUFS),LRA=动态范围(LU)
ffmpeg -v info -i in.mp3 -t 300 -af ebur128 -f null - 2>&1 | grep -A8 "Summary"
# 全片电平:RMS level(均方根电平) / Peak level(峰值) / DC offset
ffmpeg -v info -i in.mp3 -af astats -f null - 2>&1 | grep -E "RMS level|Peak level|DC offset"
# 静音段检测(底噪探针):调低阈值仍检不出静音 = AGC 把底噪抬起来了
ffmpeg -v info -i in.mp3 -af silencedetect=noise=-38dB:d=2.5 -f null - 2>&1 | grep silence
# 同一时间段截取对比电平(两文件同 offset 比 RMS,看增益差异)
ffmpeg -v info -ss 490 -t 4 -i in.mp3 -af astats -f null - 2>&1 | grep "RMS level"
# 双声道是否真立体声:astats 输出 Channel 1/2 的 RMS,几乎相同 = 同源信号非真立体声
判读规则(2026-08 实测):
- Peak > 0dB → 削波风险(前端增益过激),如飞书录音豆 1.38dB
- 动态范围 LRA < 9 LU → 强压缩;同材料 A 有静音段而 B 检不出 → B 的 AGC 抬噪
- 底噪/压缩不致命(语音主体清晰即可转写),但可分出"信号干净度"高低
- ⚠️ 声道数不等于质量:双声道若两通道 RMS 差 <0.5dB 是同源复制,对转写无额外信息
二、SRT 变体识别与统计(先识别格式再清洗)
见 scripts/analyze_srt.py(解析任意变体 → 结构统计 + 术语计数 + 锚点定位)。
已知变体(2026-08 实测,详见 references/transcription-products-2026.md):
- 飞书妙记:标准 SRT,无说话人标记 → 直接可用
- 钉钉听记:文本行首带内嵌时间戳(
00:00:02,760推动),清洗正则re.sub(r'^\d{2}:\d{2}:\d{2},\d{3}\s*', '', line) - 通义听悟:导出为纯文本段落、无任何时间戳 → 不能做字幕/时间轴定位
- 空文件排查:0 字节 SRT = 导出失败/为空,先确认再分析,别当"没内容"跳过
结构指标:块数、总字数、字/分(中文口语 200~300 正常)、片段时长中位、<1s 微块占比(碎片化,>15% 偏碎)、是否含说话人标记。
三、转写质量对照方法(没有金标准也能比)
- 关键术语识别对照:找音频里反复出现、ASR 容易错的词(产品名/英文术语/人名,如 "Claude Code"),在各家文本里数正确/错误出现次数。实测效果极佳:同一句话钉钉 8/8 全对、飞书 6 次全错(Clang/Clash code)——铁证级差异。
- 内容锚点对齐(跨产品时间轴不可信时):用独特短语(如"MCP"、"替我审批")在每份转写里定位时间戳,对比位置差。恒定差 = 开机时间差;非恒定差(前段错位 4-17 分钟、后段对齐)= 某台设备中途暂停/智能暂停 → 报告注明"按内容锚点对齐,非时间戳对齐"。
- 幻觉检测模式:凭空英文客服话术("Okay, thank you very much...")、无意义字母组合("AN""DK")、脏话误听("我他妈")、术语变乱码(skill→sql/scale/SRO)。免费版/弱模型幻觉远多于强模型。
- 同句铁证:找同一个长句在各家的完整版并排输出(搜"大长条"这类独特词),人工一读便知高下。
- CER/字错率(要严谨数字时):抽 3-5 分钟人工精校为金标准,用
jiwer(pip 安装,WER/CER)计算;无金标准时不做,别编数字。
What ships with it
2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 12d ago First seen · 81 lines · 154 tokens per session scan A b319fc0070cd
transcription-quality-evaluation is a skill published in the GitHub repository bleakbelladonnals/asr-transcription-skills (9 stars, last pushed 26d ago), licensed MIT. It adds 154 tokens to every session and 2,197 once invoked, about $0.0008 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
instrument-data-to-allotrope
Convert laboratory instrument output files (PDF, CSV, Excel, TXT) to Allotrope Simple Model (ASM) JSON format or flattened 2D CSV. Use this skill when scientists need to standardize instrument data for LIMS systems, data lakes, or downstream analysis. Supports auto-detection of instrument types. Outputs include full…
matlab
Build, review, migrate, and safely plan MATLAB or GNU Octave numerical workflows, including arrays, tabular/time data, tests, projects, graphics, MAT files, and explicit Python interoperability.
exploratory-data-analysis
Perform bounded, local exploratory analysis of explicitly supported scientific files. Use for redacted CSV/TSV/JSON profiles; optional NumPy, HDF5, FASTA/FASTQ, and basic image metadata inspection; missingness/leakage audits; outlier and transformation sensitivity; and rigorous EDA report scaffolds. Other domain…
phylogenetics
Build and analyze phylogenetic trees using MAFFT (multiple alignment), IQ-TREE 2 (maximum likelihood), and FastTree (fast NJ/ML). Visualize with ETE3 or FigTree. For evolutionary analysis, microbial genomics, viral phylodynamics, protein family analysis, and molecular clock studies.
research-engineer
An uncompromising Academic Research Engineer. Operates with absolute scientific rigor, objective criticism, and zero flair. Focuses on theoretical correctness, formal verification, and optimal implementation across any required technology.
mapping-to-snomed
Maps clinical concept spans extracted by OpenMed to SNOMED CT concepts through a USER-SUPPLIED terminology server (the user's own Ontoserver, Snowstorm, or UMLS/UTS), never a bundled vocabulary. Use when the user wants to code findings, disorders, procedures, body structures, or substances to SNOMED CT, run an ECL…