LumenVox adds German, additional enhancements with Version 15.1 of its speech engine.
Speech technology provider LumenVox has released LumenVox 15.1, the latest iteration of its suite of products. The new version comes hot on the heels of LumenVox's last numbered release, which debuted in October. Version 15.1 adds German language support to the Automated Speech Recognition (ASR) Library, along with performance enhancements and usability improvements.
The primary focus of LumenVox's current build has been the addition of German to the ASR product. LumenVox CEO Ed Miller notes that its addition was the longest in production, taking three out of the four months that transpired between version 15.0 and this latest release.
Of LumenVox's eight other supported languages, six have undergone improvements to pronunciation generation and processing. They are Canadian French, U.K. English, Brazilian Portuguese, Mexican Spanish, Columbian Spanish, and Indian English. Many of LumenVox's ASR languages, which already report confidence values (measures of system certitude for transcription) in the high 90 percentiles, are still seeing gains of several percentage points.
The latest version of LumenVox also places emphasis on load balancing for multitenant media servers for ASR and text-to-speech (TTS). The new load distribution enhancements ensure that requests to LumenVox media servers receive optimum resources. "As LumenVox has expanded and become more and more attractive to enterprises, cloud and multitenant processing have become a huge focus," notes Miller, who mentions that the company, which has been growing steadily for 15 years, saw a 20 percent increase in business across all its products in 2016, making resource management a priority.
LumenVox diagnostic tools have also been upgraded. "The diagnostic dashboard has been redesigned and now reflects new items clients can track" Miller says. These new items include the aforementioned multitenant media servers, allowing clients to tag logs and call records with identifying metadata to track their own requests through their hub media server. Additionally, clients can track security weaknesses, communications between the tenant and the media server, and the consumption of system resources.
Finally, LumenVox has made usability improvements to the Speech Tuner system. The system allows clients to adjust how other LumenVox products process utterances. "The newest version of the Speech Tuner is more consumer-esque," Miller says. "The wizard has been enhanced and will automatically recommend changes to better the performance of our other applications."
The new Tuner still requires a minimal coding investment from users, but is now more streamlined and requires less heavy-lifting from clients, he maintains.
Miller promised even greater focus on the Speech Tuner in future versions. "There's going to be a lot more," he says.