Set up voice input

Voice input turns your speech into text in the message box. The first time you use it, DataMind Oda may need to prepare its recognition models on the computer, and you choose the service and language it uses. How recording and dictation work is covered in Voice input.

When Oda asks you to set up voice input

Voice input is part of a plugin you switch on first; Plugins and bundles covers that.

After voice input is switched on, Oda checks the computer for speech models. While the models are not prepared, trying to record shows the setup prompt Set up voice input before recording, which explains: Voice input is enabled. First use requires downloading and preparing local recognition models. …

• Go to setup opens the preparation details. It does not start a download; nothing is downloaded until you press the download button there.
• Later closes the prompt and leaves voice input switched on. You can prepare the models later from the plugin's details.

Until the models are ready, the microphone cannot record, and its tooltip says Prepare recognition in the voice plugin settings first. If intact models are already on the computer, Oda skips the prompt and downloads nothing again.

Prepare the models

1. Open the preparation details. Choose Go to setup in the prompt, or open the voice-input plugin's details from the Plugins page.
2. Expand the card. It starts collapsed; expanded, it shows Preparation steps with the individual steps.
3. Press Download and prepare.
4. Follow the steps as they run: Check local resources, Prepare recognition model, Prepare voice detection model, Verify model files, and Load with the model's name.
5. Watch the statuses. Each step shows one of Not started, In progress, Completed, Failed or Cancelled; while the active step works, its line counts the wait, for example Waiting 5 seconds, and the download shows its progress.
6. Wait until the summary reports that the models are ready: Local resources are prepared. Recording wakes the recognizer. or Local speech recognition is ready.. With a cloud service it reads Cloud speech recognition is ready.

Before you start, the card shows three estimates:

• Disk space: about 5 GB for the local models.
• Memory: about 2 GB once the models are loaded.
• First setup: how long the download and preparation are likely to take.

The card also says: Local models will be downloaded to the computer running DataMind Oda. No Python or compiler is required.

Cancel preparation stops a running preparation. It is hidden while the recognizer is waking and disabled while the cancellation is in progress. If a preparation fails, start it again from the card with Retry preparation.

Choose the service and language

The card's settings choose how Oda recognises speech:

• Recognition service chooses the service that recognises your speech. A local service runs on the computer running DataMind Oda; a cloud service is reached over the internet.
• Recognition language lists the languages to listen for: Detect automatically, Chinese, English, Cantonese, Japanese and Korean. The list follows the service, so changing the service can change which languages you can pick.

Your choices are saved as you set them and persist for later sessions. Where the audio goes depends on the service:

• With the local service, audio is recognised on the computer running DataMind Oda, and the first setup downloads the models to that computer.
• With a cloud service, the card says: Audio will be sent to the selected cloud service.
• Local means the computer running DataMind Oda, which can be a different computer from the one whose browser you use.

When preparation fails

If the preparation stops with a failure, the card shows the message, a short advice line for the cause, and two values:

• Download source: <source> names where the models were being fetched from.
• Error code: <code> names the error the download hit.

The advice matches the cause. The card distinguishes a network problem, a name-resolution problem, a timeout, a certificate problem, an HTTP error from the download host, a failed file-integrity check, and not enough storage space; anything else is reported with the message and code as the details. For a name-resolution problem, the advice points at the DNS and proxy settings of the computer running DataMind Oda.

Start the preparation again from the same card. If it keeps failing, contact your DataMind OS administrator and include the message, the download source and the error code.

Failures that come from saving a setting, such as switching the service, show on the card as Speech recognition failed: <message>.

Questions, answered

Why does Oda ask me to set up voice input?

The local recognition models are not prepared yet. Oda checks the computer after voice input is switched on and asks you to prepare the models only when they are missing; the prompt stays until they are ready. When intact models are already there, the prompt does not appear and nothing is downloaded again.

Does Go to setup start the download?

No. It only opens the preparation details. Nothing is downloaded until you press Download and prepare there.

What does Later do?

It closes the prompt and leaves voice input switched on. Prepare the models later from the voice-input details on the Plugins page; until then the microphone tells you to prepare the models first.

How much space and memory does the setup need?

The card estimates Disk space, Memory and First setup before the download starts. For the local models, plan for about 5 GB of disk space and about 2 GB of memory once they are loaded. The figures depend on the service.

Do I need Python or a compiler?

No. The card says the local models are downloaded to the computer running DataMind Oda, and no Python or compiler is required.

Which recognition languages can I choose?

Detect automatically, Chinese, English, Cantonese, Japanese and Korean. The list follows the selected service.

Is my voice sent to a cloud service?

It depends on the service you choose. With the local service, audio is recognised on the computer running DataMind Oda. With a cloud service, the card says Audio will be sent to the selected cloud service. Note that local here means the computer running DataMind Oda, which can be a different computer from the one whose browser you use.

What do the step statuses mean?

Not started is waiting its turn, In progress is running now, Completed is done, Failed stopped with an error the card explains, and Cancelled was stopped by Cancel preparation.

What does Waiting 5 seconds mean?

The active step counts how long it has been waiting, for example while the download runs or the model loads. It keeps counting while the step works, and there is nothing to do.

The download failed. What do Download source and Error code tell me?

They name where the models were being fetched from and the error the download hit. Together with the advice line they are the details to share when you ask for help. Start the preparation again from the card; if it keeps failing, contact your DataMind OS administrator and include the message.

I cancelled the preparation by mistake. Can I start it again?

Yes. Open the preparation details and start it again from the card.