Automatically detect audio language with the Speech Language Detection Container

We are excited to announce the release of the Speech Language Detection Container for Public Preview!

The Speech Language Detection feature is used to determine the most likely language match for a given audio where the language is not already known. By doing so, it unlocks our Speech-to-Text service to a vast number of scenarios and helps eliminate the language barrier.

Since the release of Speech Language Detection as an online service on Azure Cognitive Services, we have watched you enable new scenarios along with our Speech-to-Text and Translation services that open new doors for productivity and accessibility. Multilingual meetings, call center conversations, voicemails, and video streams can now capture every word for captioning and analytical insights. Spoken machine translation can automatically determine the source language without manual selection.  And recommendation systems can better promote video content users can understand.

Control over your data

Perhaps you have wanted to explore with the Speech Language Detection feature before but were limited because of data restrictions. This could be due to data regulations, not wanting to or being able to load all your data into the cloud. By using the container version, you can now use the Speech Language Detection feature with complete control over your data.

Control over your throughput

Have you had to deal with a weak network connection or disconnected environments leading to high latency? With the container, you can scale for high throughput, low latency, requirements by enabling Cognitive Services to run in Azure Kubernetes Service physically close to your application logic and data.

Portable architecture

You can create a portable application architecture that can be deployed in the cloud, on-premises and the edge. This allows you the flexibility to add or remove containers very easily and adapt for what your project needs.

How It Works

The process is quite simple:

1. Client runs container with audio file or stream
2. Container receives the request
3. Container runs language detection model
4. Container returns to client the highest language match

Conceptually depicted below:

The result to the client will ultimately look like this:

Get started with installing the Speech Language Detection container

Learn more about how to download and run Speech service containers, including Speech Language Detection by visiting our documentation. Let us know your thoughts or new features you would like to see on this container!

For more information you can also explore:

• Automatic language detection for speech to text
• Speech service containers frequently asked questions (FAQ)
• Overall Speech Service documentation
• Containers overall documentation in Speech Service
• Container support in Azure Cognitive Services
• Docker Hub landing page