# Speech model is missing (/troubleshooting/speech-model-missing)

Use this page when Settings shows **Speech model is missing**, **Speech model won't load** or **Cloud speech isn't set up**, or a dictation tells you to open **Speech**. ScribeKey needs a working speech model on the phone, or a cloud speech provider, to turn audio into text. Until one works, nothing you say becomes text.

## Speech model is missing [#speech-model-is-missing]

ScribeKey is set to on-device speech, and no speech model is downloaded.

1. Open **Settings**. Tap **Choose model** on the card at the top.
2. On **Speech models**, tap **Download** on a model. ScribeKey suggests one for your phone.
3. Wait for the download to finish. You can leave the app. A notification shows its progress.
4. Tap **Set active** on the model, if it is not already in use.

The card at the top of Settings is gone. Your next dictation becomes text.

If the download shows **Waiting for Wi-Fi**, tap **Use mobile data**, or connect to Wi-Fi. To stop waiting for Wi-Fi in future, turn off [**Download models on Wi-Fi only**](/settings/storage#storage.download_wifi_only).

If a download shows **Download stopped. Try again.**, tap **Resume** or start it again.

## Speech model won't load [#speech-model-wont-load]

The model is downloaded but did not start the last time ScribeKey used it.

1. Open **Settings**. Tap **Open speech** on the card at the top.
2. On **Speech models**, choose another model, or delete the model and download it again.

The card goes after the next dictation that works.

## Cloud speech isn't set up [#cloud-speech-isnt-set-up]

ScribeKey is set to cloud speech, and the connection has no key, endpoint or model.

1. Open **Settings**. Tap **Open speech** on the card at the top.
2. Either tap **On device** to use a speech model, or finish the connection as in [Use your own cloud provider](/guides/use-your-own-cloud-provider).

> **Note:** A dictation recorded while the model was missing keeps its audio. Once speech works, open the session and tap **Retry transcription**.

## If this does not fix it [#if-this-does-not-fix-it]

[Choose a speech model](/guides/choose-a-speech-model) explains the choice of model. If a model still will not load, open **Settings**, then **Help & feedback**, and tap [**Report a problem**](/settings/help-feedback#help.report_problem). The report names the model and the error.

## Email support [#email-support]

Write to [support@scribekey.app](mailto:support@scribekey.app). Say which model you chose and your phone's make and model.