Skip to main content

🚀 Batch API

Batch API V2 released, check out our guide and try it out!

Notable changes:

  • Improved support for working with multiple audio files
  • Flatten ProcessAudioSource message by splitting processing_config into tasks, config and output
  • Enable specifying text output formats, lexical or display
  • Add support for translation Enterprise feature
  • Improve documentation

Batch API V1 will still be available, but it's being deprecated.

🔧 Maintenance

When creating a new application, it's now deployed directly and is ready for use. This means you no longer need to remember to press Deploy in Speechly Dashboard, or run the deploy command with Speechly CLI, after creating an application!

📝 Docs

Added a comprehensive Features table from where you can see all Speechly features and their availability.

🖥 CLI

Version 0.11.5 released, which improves the download command. Previously it would clear the output directory. Now it just downloads the latest configuration, overwriting any files with the same filename. All other files in the output directory are left as is.

🖥 CLI

Version 0.11.4 released.

  • Improve on-device transcription
  • Improve first time usage

👀 Demos

New transcription demo available. The demo accurately transcribes both live and pre-recorded audio using our off-the-shelf large-highaccuracy speech recognition model.

🔮 Dashboard

  • A lot of under the hood maintenance, hopefully the Dashboard is now even faster and smoother.
  • Fix a bug where the lookup editor would be empty

💎 ASR improvements

Our newest generation of ASR models using conformer transducer architecture are now in public beta for all users. The new architecture provides significant accuracy gains in word error rate.

🖥 CLI

Version 0.11.2 released, notable changes:

🔮 Dashboard

  • Improved application empty state and onboarding views.
  • Fix a bug where a previously deployed application that is currently in training couldn't be previewed.
  • Fixes to local storage handling.

🚀 Streaming & Batch API

As of next week, we're switching to using lowercase letters in the transcript from our APIs. The intent and entities will be unaffected by this, but after the change the contents of the segment.words list would be in lowercase.

If you are currently relying on the transcripts being in uppercase for some feature, please make sure to update that logic. For most users this won't be an issue, though.

📝 Docs

Today we are extremely happy to finally announce version 2.0 of our documentation! 🥳️

  • We rewrote and restructured most of our content in order to make things clearer and improve the developer experience.
  • The site has a new look and feel with better mobile and dark mode support.
  • Even the search is now better, we hope you'll find what you are looking for :)

🖥 CLI

Version 0.11.0 released, notable changes:

  • Transcribe audio files using Speechly Cloud or On-device with the transcribe command.
  • Evaluate both ASR and NLU accuracy of your application using the evaluate asr and evaluate nlu sub-commands.

🔮 Dashboard

A lot of changes to application views:

  • Application views were re-organized and some of them were renamed to better describe their purpose.
  • Download SDKs for Speechly On-device from the Integrate page
  • A new design for the Training data (previously Configure) view. The editor is now full height with a sticky sidebar, so less scrolling to find your entities and intents.
  • Templates were removed when creating a new application
  • A bunch of small, yet important, UI improvements

📦 Speechly Decoder

Version 1.1.0 of the Speechly Decoder library released for iOS, Android and Unity.

🖥 CLI

Version 0.10.0 released, which adds support for downloading on-device model bundle files using the download command.

🔮 Dashboard

Add support for downloading on-device model bundle files.

📦 Speechly Decoder

Version 1.0.0 of the Speechly Decoder library released for iOS, Android and Unity.

🖥 CLI

Version 0.9.0 released, which unifies the deploy and download commands. Now both work in similarly and the --out flag is no longer needed.

🔮 Dashboard

See who created those API tokens and remove users from a project.

📦 Browser client

Version 2.3.0 released which adds voice activity detector (VAD) feature for building experiences without a microphone button.

📦 Unity client library

With the Speechy client library for Unity, developers can now easily add Voice UIs as a Feature to their experiences built with Unity. Read our blog post or check out the docs to learn more.

🖥 CLI

Version 0.8.2 released.

🔮 Dashboard

A bunch of small UI tweaks in Project Settings and Configure views to enhance the usability of the Dashboard.

🔮 Dashboard

Added advanced settings such as silence triggered segmentation, rule-based NLU and others to tweak your application eve further. Find them by going to Application > Settings.

🔧 Fixed

Several under-the-hood improvements to make your developing experience faster and smoother.

🔮 Dashboard

Intents and entities now have a unified look throughout the Dashboard. Blue for entity and green for intent. Happy developing!