> ## Documentation Index
> Fetch the complete documentation index at: https://comfytv.org/llms.txt
> Use this file to discover all available pages before exploring further.
# Audio Picker
> Pick **one** audio track out of a pool of generated clips to send downstream—the audio counterpart of Image Picker.
## What this node does
**Audio Picker** is a **Compose** stage (`ComfyTV/Compose`). Upstream stages that produce audio—**Music Stage**, **Speech Stage**, or any node whose runs pile up multiple takes—feed into this node. The node body shows the accumulated tracks; whichever one you click becomes the sole **`COMFYTV_AUDIO`** on the **audio** output.
There is **no ▶ Run**. Clicking a track updates the hidden `selected_index` and re-emits the output snapshot instantly, so downstream stages immediately see your pick.
Media flows as ComfyTV project snapshots, not native ComfyUI tensors. To hand a picked track to native ComfyUI audio nodes, insert a **Bridge** — see [bridges.md](https://github.com/jtydhr88/ComfyTV/blob/main/docs/bridges.md).
## When to use it
* A **Music Stage** returns several takes across multiple Runs and you want to audition and keep one.
* You generated a few **Speech** reads of the same line and need to choose the best delivery.
* You want to **accumulate** takes across Runs into one pool and pick without losing earlier ones.
## Parameters
### batch (upstream)
The wired input, of type `COMFYTV_AUDIO` and optional. Each generated track that arrives here is appended to the pool. Wire it from a Music Stage or Speech Stage `audio` output.
### selected\_index (hidden)
1-based index of the track to extract (default `1`, range `1`–`999`). You never type this—it is set by **clicking a track** in the node body.
### pool (hidden)
Accumulated audio pool, stored as JSON on the node. The UI appends new upstream tracks (deduped by URL) so they survive regeneration and disconnect, and empties it via the **Clear** button. When the pool is non-empty it is the source; otherwise the live `batch` input is used.
### project\_id / parent\_output\_id (internal)
Hidden; maintained by the **Project** node and graph wiring. You rarely touch these.
## Outputs
| Output | Type | Meaning |
| --------- | --------------- | -------------------------------------------------------- |
| **audio** | `COMFYTV_AUDIO` | The single selected track snapshot for downstream stages |
## Tips
* Nothing appears until you actually **click** a track—selection, not hover, sets the output.
* Pool cluttered with old takes? Use **Clear** in the node body, or drop in a fresh Audio Picker.
* The pool persists selections and takes with the project, which native in-memory audio tensors cannot—use a Bridge only when crossing to native nodes.
## Related nodes
* **Music Stage** — common upstream; generates instrumental / vocal tracks.
* **Speech Stage** — text-to-speech upstream.
* **Image Picker** / **Video Picker** — same pick-one-from-a-pool pattern for other media.
* **Load Audio from Asset** — pull an existing track from the project asset library instead of picking from a pool.