# Audio Parser: Rename Sound Files From the Spoken Slate

Source: https://storiara.com/tools/audio-parser
Last updated: 2026-09-14
Author: Nick Harty, Storiara

> The audio parser listens to uploaded audio files for a spoken slate, such as "scene 42, shot 3, take 2", and pulls out the scene, shot, and take numbers. You can check and edit each result, add an optional suffix, and download the files renamed as Scene_42_Shot_3_Take_2, one at a time or together in a ZIP.

## Processing a batch

Drag audio files onto the upload area or click **Choose Files**. Several files can go in at once. Each one shows a progress card that moves from Compressing to Processing to Completed, or Error with a reason.

Completed files appear in a results list in the order you added them. Each result shows the original file name, the scene, shot, and take the parser found, and a suffix field. From there you can:

- press **Play** to listen to the file in the browser and check the slate against the result
- click any scene, shot, take, or suffix value to edit it
- press **Download** to save that file with its new name
- press **Download All** to get every result in one ZIP named with today's date (the button appears once more than one file has finished)

You can keep adding files after the first batch. New results are added to the same list.

## What happens to each file

1. The browser checks that the file is audio and under 200 MB.
2. It decodes the audio, mixes all channels down to mono, and drops the sample rate to 16 kHz or lower, then packages that as a WAV. If decoding fails, the original file is sent instead.
3. The reduced copy goes to Storiara's server, which asks Google's Gemini model for the scene, shot, and take numbers only, as numbers and letters like 1, 12, or 42A. Anything it can't find comes back as Unknown.
4. The download uses your original file, at its original quality and extension, under the new name.

The name follows one pattern:

`Scene_<scene>_Shot_<shot>_Take_<take>.<extension>`

with `_<suffix>` added before the extension if you typed one.

## Worked example

A mixer's recorder names its files by number, so the dailies folder holds `T012.WAV`, `T013.WAV`, and `T014.WAV`. On each file the 2nd AC or the mixer said the slate out loud before the take.

| File | Spoken slate | Parser result | Download name |
|---|---|---|---|
| T012.WAV | "Scene 42A, shot 3, take 1" | 42A / 3 / 1 | Scene_42A_Shot_3_Take_1.WAV |
| T013.WAV | "Take 2" | Unknown / Unknown / 2 | edit to 42A and 3 first |
| T014.WAV | "Scene 42A, shot 3, take 3" | 42A / 3 / 3 | Scene_42A_Shot_3_Take_3_boom.WAV, after typing "boom" as the suffix |

T013 shows the common failure. Only the take number was spoken, so the parser had nothing else to go on. Fix the two Unknown fields by hand before you download, or the file will be named `Scene_Unknown_Shot_Unknown_Take_2.WAV`.

## Where it gets things wrong

- Metadata is ignored. Many production recorders write scene, take, and notes into the file itself. If yours does, that metadata is more reliable than a spoken slate.
- Phonetic letters. A mixer who says "forty-two apple" or "forty-two A" may or may not get back 42A. Listen and check A-scenes and pickups.
- One slate per file. If a file holds several takes, the parser returns one set of numbers.
- Noisy slates. Wind, generator noise, or someone talking over the slate can lead to wrong or Unknown results.
- Shot numbering. Productions that slate setups as letters (42A, 42B) have no separate shot number to say, so the shot field may come back Unknown. Edit it or type something consistent, such as the setup letter.
- Unsupported characters. The name uses whatever you type in the fields, so avoid slashes and other characters your file system won't accept.
- Limits. Files over 200 MB are rejected, and the 200-request daily count is per IP address, so a whole office sharing one connection shares the count.
- Files that fail don't appear in the results list, and results aren't saved if you leave the page.

## Where this fits in post

Renamed files are easier for the [assistant editor](https://storiara.com/roles/assistant-editor) to sync against picture and match to the [sound report](https://storiara.com/glossary/sound-report). They don't replace either. The sound report from the [production sound mixer](https://storiara.com/roles/production-sound-mixer) is still the record of what each file holds, including [room tone](https://storiara.com/glossary/room-tone), wild lines, and which takes were circled. For how numbering works on the slate in the first place, see [slate](https://storiara.com/glossary/slate), [scene numbering rules](https://storiara.com/blog/scene-numbering-rules), and [take](https://storiara.com/glossary/take).

## Frequently asked questions

### What audio formats does the parser accept?

Any file your browser identifies as audio, including WAV, MP3, and M4A, up to 200 MB per file. If one file in a batch isn't audio or is too large, the whole batch is stopped with an error before anything uploads.

### Does it read the scene and take from WAV metadata?

No. It only listens to the sound in the file. Scene and take fields that a recorder writes into the file's metadata are ignored, so a take without a spoken slate comes back as Unknown.

### Is my audio uploaded anywhere?

Yes. A reduced copy of each file (mono, at most 16 kHz) is sent to Storiara's server, which passes it to Google's Gemini model to find the numbers. The original file stays in your browser and is the one you download.

### Is there a limit on how many files I can process?

Each file counts as one request, and the limit is 200 requests per IP address in a 24-hour window. After the first file, the tool shows how many you've used and when the count resets.