Bug: Unable to select AssemblyAI transcription provider in Podcast Content Studio

[teethatkamsai]

Description

After pulling the latest changes (including my previous PR that added support for AssemblyAI), I'm unable to select AssemblyAI as the transcription provider in the Podcast Content Studio.

Currently, the UI only shows:

• Paste transcript
• Auto (Whisper)

There is no option to choose AssemblyAI, even though the functionality was introduced in the previous pull request.

Expected Behavior

The transcript section should allow users to choose between available transcription providers, for example:

• Whisper
• AssemblyAI

If AssemblyAI is selected, the application should use the configured AssemblyAI API for transcription.

If Whisper is selected, the existing Whisper workflow should remain unchanged.

Current Behavior

Only the Whisper option is available.

There is no way to switch to AssemblyAI from the UI.

Screenshot

(Attach the screenshot below)

Suggested Solution

Add a transcription provider selector to the Transcript section.

Example:

Transcript Provider

○ Whisper
○ AssemblyAI

or a dropdown:

Transcription Provider
[ Whisper ▼ ]

Behavior:

• When Whisper is selected:

  • Use the existing Whisper transcription pipeline.

• When AssemblyAI is selected:

  • Use the configured AssemblyAI API endpoint and settings.

• Preserve the current Whisper experience as the default if no provider is selected.

• Expose the provider selection in both the UI and CLI so users can consistently choose their preferred transcription engine.

Environment

• Latest main
• Windows
• Podcast Content Studio

---

[nmbrthirteen]

Thanks for the report. The AssemblyAI selector actually shipped in v2.3.0 (commit 5f27df4), which released about 30 minutes before this issue was filed, so the build you pulled predates it.

The tell is your screenshot: the tab reads "Auto (Whisper)", which was the label before the selector landed. On v2.3.0+ that tab is just "Auto" and contains an Engine dropdown (Whisper / AssemblyAI) plus an API-key field.

What's already available on latest:

• UI (EpisodeWorkspace.jsx) — Auto tab → Engine dropdown → pick AssemblyAI → enter your key (or set ASSEMBLYAI_API_KEY in Config).
• CLI — podcli process and podcli studio both accept --engine whisper-py|whispercpp|assemblyai and --assemblyai-api-key (prefer the ASSEMBLYAI_API_KEY env var so the key doesn't show up in process listings).

Please update to v2.3.1 and pull latest main:

curl -fsSL https://podcli.com/install.sh | sh

One genuine gap I did fix on top of this: the engine choice was resetting on reload. #50 persists it with the preset config (same as the Whisper model), so a saved show keeps its preferred engine. The API key stays in the global .env, not in presets.

Closing as resolved by v2.3.0/v2.3.1, but reopen if the selector still doesn't show after updating.