Documentation Center

Connecting to a Nuance server for audio transcription

As of SDL Machine Translation Edge 8.0, you can connect your SDL Machine Translation Edge instance to a Nuance server you purchased separately. If you have an audio file (.wav) you want to translate, the file must first be transcribed by Nuance before SDL Machine Translation Edge can translate it.

Procedure

  1. Install the Nuance ASR on a server which SDL Machine Translation Edge can access.
  2. Access the SDL Machine Translation Edge master host.
  3. In the main-config.json file, add the following section, at the bottom of the file, under the highest parent node:
    Nuance information for the main-config.json file
    Speed values can be: 'accurate' (1 x real time), 'fast' (3 x real time), 'warp' (10 x real time) with cost of quality being exchanged for speed.
    Nuance example for setting host details
  4. Restart your SDL Machine Translation Edge instance (if it is running).

What to do next

You can now translate audio files from formats that Nuance accepts into text translations:
  • In the Web GUI, drag and drop an audio file via the Quick Translate page or use the file browser. Only .wav files are currently supported.
  • In the API, follow the steps for submitting a file for translation, but specify the audio 'mimetype' as the InputFormat. Remember to base64 encode the input text.