Speech Transcription Server

Overview

GL’s Speech Transcription Server (STS) is a PC-based automated Speech-to-Text conversion application. Among numerous applications, the Speech Transcription Server can be used for confirming voice prompts (announcements) and aid testing Interactive Voice Response (IVR) systems as well as voice transportation over any network. Network providers use the application to record the voice prompts associated with the IVR, perform a Speech to Text conversion on the recording to confirm the prompt was proper (based on what the prompt should be), and thus confirming their IVR functioning. The application can be used to verify network quality as well as effect of different codecs on the speech quality.

The STS utility includes a text-to-speech feature that allows automated, on-the-fly generation of audio files from user-defined text. The speech synthesizer provides options to configure languages (over 50 supported) and save audio files in PCM/WAV format. Text can be manually entered to generate voice prompts in the selected audio file format. The converted audio files are automatically saved in the specified directory path.

The STS utility supports the speech-to-text feature, enabling seamless conversion of recorded audio into text. It supports formats such as PCM, WAV, and GLW, ensuring high transcription accuracy with confidence scores ranging from 0 to 1. This functionality also processes audio in real-time and stores results for future analysis or use.

This application works with GL’s Audio File Conversion Utility which supports various industry standard codecs. This application can be good companion for GL's MAPS™, and VQuad™ applications that allow transmission and recording of voice to/from audio files over many industry standards telecommunication networks such as VoLTE, UMTS, GSM, FXO, FXS, 4-Wire, ISDN, SS7, etc.

By incorporating GL’s Speech Transcription Server, MAPS™ IVR allows users to automate precise IVR tree traversal configurations through MAPS™ profiles and test for pass/fail conditions. IVR menus are divided into configurable prompts where certain keywords are expected. MAPS record each prompt for speech-to-text transcription and analyzes the transcribed text.

The Speech-to-Text application can also supplement GL's Voice Quality Testing solution providing a more passive method to verify good quality audio. Using the VQT intrusive method for Voice Quality, the customer needs to send a pre-defined voice sample from near to far-side, record at the far-side and analyze based on the file sent from near-side. This is a very accurate method for generating a Voice Quality measurement but requires equipment at both ends of the call.

Alternatively, if access to both sides of the call is not available, the Speech-to-Text method can simply record a pre-specified sentence at the far-end and provide a Certainty score of the recorded Text based on what was expected. The Speech to Text conversion can confirm if the received audio, for instance a customer speaking into their phone, matches what was expected which would confirm there was relatively good audio on the network.

Main Features

• Ability to convert PCM or WAV or GLW files into text format
• Monitor single or multiple folders containing audio files for automatic transcription
• Each monitored folder can be configured for language and audio format
• Full automation using VQuad™ scripting
• Transcribe up-to 30 seconds of speech files into text
• Provides options to transcribe speech to text manually for a single PCM or WAV or GLW file using Manual Transcription option
• Accurate analysis of transcribed text with quality (Pass/Fail) scores
• Transcribe recorded short speech files into text; there is a maximum limit of 30 seconds for any single utterance
• Cloud-based processing provides accurate translations
• Concurrent transcription of up to 30 voice files
• Supports multi-languages such as U.S./U.K. English, French, German, Italian, Japanese and more
• Easy to access transcribed text via file, API or database
• Base Software includes up to 420 hours of audio transcription per year validity can be extended with annual support contract
• REST API support for transcription request and transcript retrieval
• Automated on-the-fly generation of audio files from user-defined text
• Ability to convert text files into PCM (u-law, a-law, PCM16-8 kHz, PCM16-16 kHz, PCM16-48 kHz) or WAV (u-law, a-law, WAV16-8 kHz, WAV16-16 kHz, WAV16-48 kHz) speech file formats