Speech Data Collection

Build better natural language processing, understanding, and automatic speech recognition solutions with human-annotated speech data in over 180 languages and dialects

Use case

When training your automatic speech recognition (ASR) system, data quality and quantity are both critical. You need high-quality language data to ensure your system can understand and respond to human speech in a variety of environments and contexts. You also need large volumes of data to train your machine learning model effectively and produce the right degree of situation diversity and solution accuracy. It’s important to collect natural language utterances (NLUs), which help train and test applications to recognize the nuances of human speech.

Our approach

Appen’s end-to-end speech data collection service delivers efficiency and quality, even when running multiple large-scale speech collection programs in parallel. Our services include natural language utterance collection through our smartphone app, as well as centralized on-site recordings in a wide range of acoustic environments. Our speech collection services cover a variety of types, including telephony, embedded device, single/multi-speaker, prompt variation, speech modality, text corpora, and other resources.

Our speech data collection services offer you:

  • Detailed linguistic and cultural research
  • Script preparation and localization
  • Crowdsourcing of native speakers
  • Local and remote speech recording
  • Transcription and annotation of collected data
  • Transcription and annotation of collected data
  • Quality assurance and project management
  • Lexicon entries matching database contents