Download podcast audio from RSS feeds and transcribe to text using AuralWise API. This skill should be used when the user wants to download podcast episodes, convert podcast audio to text transcripts, or batch-process a podcast library for searchable content. Triggers include downloading podcasts, podcast transcription, audio-to-text conversion, RSS feed downloading, or any request involving podcast audio acquisition and speech-to-text conversion. Covers RSS feed discovery, audio downloading, AuralWise API transcription, and AI-generated content overviews with book references, key concepts, and searchable keywords.
Automate the full pipeline of podcast content acquisition: discover RSS feed, download all episodes, transcribe audio to text via AuralWise API, and generate structured content overviews for each episode. Output includes audio files, Show Notes, full transcripts (.txt / .md / .srt), and AI-generated episode summaries with extracted books, concepts, and keywords.
Workflow
Step 1: Discover RSS Feed
Determine the podcast's RSS Feed URL. Use one of these methods in priority order:
User provides URL directly — proceed to Step 2.
Apple iTunes Search API — search by podcast name:
text
GET https://itunes.apple.com/search?term={podcast_name}&media=podcast&limit=5
Extract feedUrl from results. See references/rss_discovery.md for details.
Platform-specific patterns — e.g., Ximalaya albums use https://www.ximalaya.com/album/{id}.xml.
Verify the RSS Feed is accessible by fetching it and confirming episode count.
Step 2: Configure AuralWise API Key
The transcription service requires an AuralWise API Key. Handle this interactively:
Check if API Key is already configured — look for a .env file in the working directory containing AURALWISE_API_KEY=asr_.... If found and valid, skip to Step 3.
{output_dir}/
├── 音频/ # Downloaded audio files (.m4a)
├── 文稿/ # Show Notes (.md)
├── 文字稿/ # Transcripts
│ ├── EP01-xxx.txt # Plain text
│ ├── 带时间戳_EP01-xxx.md # Markdown with timestamps
│ └── EP01-xxx.srt # SRT subtitles
├── pipeline_state.json # Resume state
└── pipeline.log # Run log
Cost estimation: AuralWise optimize tier = 0.27 CNY/hour. For a podcast with ~170 hours of audio, total cost is ~46 CNY.
Step 4: Generate Content Overviews
After transcripts are generated, create a structured overview Markdown file for each episode. This is an AI-generated summary — read the .txt transcript and produce a 概览_{filename}.md file in the same 文字稿/ directory.
Read the full .txt transcript file to understand the episode content
Extract all book titles mentioned (look for patterns like "《书名》", "the book", author names)
Identify 5-15 core concepts discussed in depth
Pull 2-5 memorable quotes verbatim from the transcript
Generate 10-20 searchable keywords wrapped in backticks for easy full-text search
Keep summaries concise but information-dense — the goal is fast retrieval, not replacement of the original transcript
Save as 概览_{original_txt_filename_without_ext}.md in the same 文字稿/ directory
Step 5: Resume / Batch Processing
The pipeline supports 断点续传 (resume). If interrupted:
Re-run the same command — completed episodes are automatically skipped
State is tracked in pipeline_state.json
If AuralWise balance runs out mid-run, remaining episodes are downloaded but transcription is skipped
After recharging, re-run to process remaining episodes
Important Notes
Audio URL accessibility: AuralWise transcribes by fetching the audio URL directly — no need to download then upload. Verify URLs are publicly accessible.
Ximalaya feeds: Audio URLs from Ximalaya RSS feeds are publicly accessible and work directly with AuralWise.
Xiaoyuzhou (小宇宙): Their API requires login tokens. Use the iTunes Search API to find the same podcast's RSS feed on Ximalaya or Apple Podcasts instead.
Rate limiting: AuralWise has a default concurrency limit of 3. The script handles 429 responses with retry.
Encoding: All output files are UTF-8 encoded. Filenames are sanitized to avoid OS-level issues.
Resources
scripts/podcast_pipeline.py — Main pipeline script (parameterized, supports any RSS feed)
scripts/.env.example — Template for API Key configuration
references/auralwise_api.md — Full AuralWise API documentation
references/rss_discovery.md — Methods for discovering podcast RSS feeds