pn-transcribe

pn-transcribe is a skill for Cursor from perniemann/pnCore. It costs 51 tokens per session (1,991 once invoked), scanned A, original, MIT.

A speech-to-text tool for turning recorded audio into written words with timing information. It can create SRT and VTT caption files, which are subtitle files used by video players.

In plain words
What is it for?
Use it to caption voiceovers, interviews, screen recordings, and other videos, including captions that appear word by word or stay synchronized with animation.
Why use it?
It removes the need to type out recordings and time each caption by hand. It also helps check whether generated narration matches the intended script.

Skill for Cursor

Written for Cursor: shipped in a Cursor plugin.

Part of the pn-core plugin — 133 skills, 19 commands, 9 agents, 1 MCP server shipped together

Good fit Use it to caption voiceovers, interviews, screen recordings, and other videos, including captions that appear word by word or stay synchronized with animation.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/perniemann/pncore/pn-transcribe
Install

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.

Any agent
npx skills add perniemann/pnCore --skill pn-transcribe
Clone the repo
git clone --depth 1 https://github.com/perniemann/pnCore

Made for: Cursor.

Or install pn-core, the plugin that ships this one along with the rest of its 133 skills, 19 commands, 9 agents, 1 MCP server.

Wrote 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.

agentmods badge for pn-transcribe

README.md
[![agentmods](https://agentmods.dev/badge/skills/perniemann/pncore/pn-transcribe/github.svg)](https://agentmods.dev/skills/perniemann/pncore/pn-transcribe)
Your own site
<a href="https://agentmods.dev/skills/perniemann/pncore/pn-transcribe"><img src="https://agentmods.dev/badge/skills/perniemann/pncore/pn-transcribe/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.

agentmods 80×15 button for pn-transcribe

Your own site · 80×15
<a href="https://agentmods.dev/skills/perniemann/pncore/pn-transcribe"><img src="https://agentmods.dev/badge/skills/perniemann/pncore/pn-transcribe.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 51 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,991 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5.1 $0.00051 $0.01991
Opus 5 $0.00026 $0.00996
Sonnet 5 $0.00010 $0.00398
Haiku 4.5 $0.00005 $0.00199

Measured 8d ago against content hash 2028057664a6, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-11, from the pricing page.

Security

Grade A, and why

pn-transcribe 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 8d 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.

packages/pn-core-mcp/content/skills/media/pn-transcribe/SKILL.md · 204 lines

How it starts

The opening of the file, as written. The whole thing — 204 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Transcription and captions (STT)

When to use

  • Generating a timestamped SRT or VTT caption track from narration audio produced by pn-tts.
  • Transcribing screen recording audio, interview recordings, or voiceover for accessibility compliance.
  • Producing word-level timestamps for caption-synced animation (bouncy captions, karaoke-style reveals).
  • Validating that AI-generated TTS matches the intended script (hallucination check).

Tool selection

Tool Model Best fit Notes
Whisper (OpenAI API) whisper-1 Quick, accurate, many languages, word timestamps response_format: verbose_json for timestamps
Whisper (local / faster-whisper) large-v3, medium Offline, batching, custom alignment faster-whisper is 4× faster than original; CTranslate2 backend
AssemblyAI Universal-2 Speaker diarization, auto-chapters, redaction Best for multi-speaker; paid per minute
ElevenLabs timestamps API When audio was generated by ElevenLabs TTS Returns character-level alignment; no separate transcription step
Gentle (forced alignment) Kaldi When you have the transcript and just need timestamps Requires known-correct transcript; aligns speech to text exactly

Default: Whisper via OpenAI API (simple, no infra) or faster-whisper local (batch, cost-sensitive, offline).

Workflow

1. Pre-process audio

Ensure clean audio before transcription — noise, music, and reverb increase error rate.

# Strip music bed, keep voice (if isolated tracks are available)
# Normalize loudness
ffmpeg -i narration.wav -af loudnorm=I=-14:TP=-1:LRA=11 narration-clean.wav

2. Transcribe with word-level timestamps

OpenAI API:

from openai import OpenAI
client = OpenAI()

with open('narration-clean.wav', 'rb') as f:
    result = client.audio.transcriptions.create(
        model='whisper-1',
        file=f,
        response_format='verbose_json',
        timestamp_granularities=['word', 'segment'],
    )

# result.words: [{'word': 'Hello', 'start': 0.0, 'end': 0.24}, ...]
# result.segments: segment-level blocks for SRT generation

Read the full file on GitHub · 204 lines

Changes

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.

  1. 8d ago First seen · 204 lines · 51 tokens per session scan A 2028057664a6

Subscribe to this mod's changes

pn-transcribe is a skill published in the GitHub repository perniemann/pnCore (0 stars, last pushed 5d ago), licensed MIT. It adds 51 tokens to every session and 1,991 once invoked, about $0.0003 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 2026-09-03.

Related

Other skills, from other repositories

typescript

TypeScript coding conventions, best practices, and patterns for writing clean, maintainable code.

genkit-ai/genkit · 20 tokens

ax-java-audio

Use when writing Java code with dev.axllm:ax for audio input/output, OpenAI Responses audio mapping, realtime event folding, and generated package audio examples.

ax-llm/ax · 40 tokens

generate-images-with-firebase-ai

Use when generating or editing images from Flutter/Dart with Firebase AI Logic and a Gemini image model (Nano Banana), making the first call work, choosing Gemini Developer API vs Vertex AI, hitting quota, billing or App Check failures, getting empty or image-only responses, sending a user photo as input, controlling…

evanca/flutter-ai-rules · 84 tokens

slides

Create and edit presentation decks as single self-contained HTML files with a live, editable preview and print-to-PDF export. Use when the user wants slides, a deck, or a presentation.

decocms/studio · 40 tokens

invokeai-image-gen

Generate images using InvokeAI's local API. Use when asked to generate, create, or make images with InvokeAI, FLUX.2 Klein, Z-Image Turbo, FLUX, or SDXL models. Supports text-to-image generation, automatic model detection, image download, and parameter selection based on model architecture.

sammcj/agentic-coding · 69 tokens

storytelling-with-data

Apply storytelling principles to create, review, and improve data visualisations and data-driven communications. Use when the message or story of a data communication (chart, dashboard, infographic, data-driven deck or report) needs structuring or review, or when structuring narrative arcs, hooks, or aha moments in…

sammcj/agentic-coding · 69 tokens