Build the Speech Translator Project
- An IoT Device (e.g. 2 Raspberry Pi 2 or 3 boards)
- A Headset e.g. the Microsoft LifeChat-3000 Headset
- A Mouse
- A router connected to the Internet
- An ethernet cable
- An HDMI montor and cables
- A micro-SD card and reader
Setup your Development PC (Required for Windows IoT Builds 15063 or greater.)
- Install the Windows ADK for Windows 10, version from here
- Ensure that "Imaging And Configuration Designer" (ICD) is selected.
- Install to the default location.
- Install the Windows IoT Core ADK Add-Ons from here
- Install to the default location.
Setup your IoT Device
Install a clean O/S to your IoT Device.
Connect your device to the router, and connect the router to the Internet.
Connect the headset and mouse to your device.
Boot your device.
Select a language and, if Wi-Fi is supported on your IoT Device, skip the Wi-Fi Configuration step.
Provide Voice Permission from Cortana when prompted. (Required for Windows 10 IoT Builds 15063 or greater)
- At the first Cortana Prompt, press OK.
- At the second Cortana Prompt, press Maybe Later.
Rename your device
- Use Powershell to connect to your device
- Change the name of your device to "speechtranslator"
- setcomputername speechtranslator
- Reboot your device.
Copy a Speech Language Package to your device. (Required for Windows 10 IoT Builds 15063 or greater)
Open a new powershell window on your desktop
From desktop map your device's disk to a local drive:
- net use t: \speechtranslator\c$ /user:\administrator p@ssw0rd)
Copy a language package from your desktop to your device.
Note: Language packages are available in this folder: c:\Program Files (x86)\Windows Kits\10\MSPackages\retail\ your-processor-type \fre
For example, to copy Spanish Mexican to an arm based device:
- copy "c:\Program Files (x86)\Windows Kits\10\MSPackages\retail\amd64\fre\Microsoft-Windows-OneCore-Microsoft-SpeechData-es-MX-Package.cab" t:
Unmap your local drive (e.g. net use /delete t:).
Apply the Speech Language Package to your device. (Required for Windows 10 IoT Builds 15063 or greater)
Note: Your device should reboot after committing each update and a "spinning gears" screen will appear until the update completes.
- applyupdate -stage c:\Microsoft-Windows-OneCore-Microsoft-SpeechData-es-MX-Package.cab
- applyupdate -commit
Wait for your device to reboot
Setup Azure with Cognitive Services
- Follow these instructions to configure your Azure account with the Cognitive Services APIs.
- After creating your account and subscribing to the Cogntive Services APIs make note of one of the subscription keys for your account.
From Azure Web Portal Select:
- "All Resources" (may appear as a 3x3 grid icon).
- Select your CognitiveServices subscription from the list.
- Under the "Cognitive Services account" menu, select "Keys".
- Make note of either Key 1 or Key 2, you will need to add this key to the sample source before rebuilding.
Setup your sample
- Download the sample from here to your local PC.
- Open the solution file in visual studio.
- Open the constantParam.cs file.
- Replace the subscriptionKey with either Key 1 or Key 2 from instructions above.
- Rebuild the solution.
- Deploy and run on your device.
- While wearing your headset, Press "Start Talk".
- Say something in English, the "Message Recognized" box should contain the spoken English phrase.
- Press "Stop Talk".
- After a moment the Translated Message should appear in the dialog box, and you should hear the spoken phrase through your headset.
- The process can be reversed, by changing the "Recognizer Language" to your installed language (e.g. Spanish) and speaking a phrase in that language.