# Changelog of Speech to Text: Audio, Video & Podcast Transcriber (`rod_analytics/media-transcriber`) Actor

- **URL**: https://apify.com/rod_analytics/media-transcriber/changelog.md
- **Full Actor documentation**: https://apify.com/rod_analytics/media-transcriber.md

## Changelog

### 0.1.8 (2026-10-05)

- Hosted Groq key: with the secret version env var `GROQ_API_KEY_HOSTED` set, runs without a buyer key use it.
  The buyer's own key always wins. Without the env var the Actor behaves as before and asks for a key.
- Hosted mode always uses `whisper-large-v3-turbo`. Other models and "Translate to English" need the buyer's key.
  Translation without a key ends SUCCEEDED with a free `translation-needs-own-key` row.
- Hosted key rate limited (wait over 60 s or a daily limit) or rejected: remaining files are skipped and free,
  the run ends SUCCEEDED with one free `hosted-key-busy` row.
- The prefill now returns a real transcript without any buyer key.
- All inputs invalid (bad URLs, blocked platforms, broken feed): one free `invalid-input` help row instead of one row per input.
- New dataset field `apiKeySource`: hosted, own or none.
- Pricing: new tiered entry. Audio minute $0.003 FREE to $0.0021 GOLD. Base price unchanged.
- Listing: title "Speech to Text: Audio, Video & Podcast Transcriber", Whisper in SEO title and description,
  README rewritten, real example input. Default memory stays 512 MB.

### 0.1.6 (2026-09-30)

- Audio minute price $0.002 to $0.003. Default memory 512 MB.

### 0.1.3 (2026-09-27)

- First public version. Bring your own Groq or OpenAI key, podcast RSS, SRT and VTT.