Transcription Services

As a world leader in the provision of high-quality transcription services, Appen can process several million speech utterances per month, in more than 80 languages.

Typically, Appen provides transcription to support:

• The creation of speech and language databases. Once speech data is captured, it is then transcribed and annotated to tag all relevant acoustic events so that it can then be used to develop software acoustic models in the languages of interest.
• Tuning for speech recognition applications. This process involves taking large volumes of spoken sound information captured in real-life situations where recognition is sub-optimal and transcribing this accurately. This facilitates tuning of the application by significantly improving the accuracy of the voice recognition.

Appen runs transcription operations in Sydney (Australia), Bangalore (India), Peshawar (Pakistan) and Naples (Italy), where it uses its proprietary tools and processes to ensure that it meets customer-specified quality levels irrespective of location. Quality assurance for all transcription projects is carried at in Sydney.

Our transcription services include:

• orthographic transcription
• acoustic transcription (where additional speaker, medium and background noises are also tagged)
• phonetic transcription
• annotation of other language data

Appen provides a unique level of expertise based on:

• the scale of our operations
• the diversity of transcription conventions to which we work
• the wide range of languages in which we have experience

Our experiences range from transcribing colloquial Arabic (a spoken language with an unstandardized written form) to processing projects with hundreds of thousands of utterances in just weeks.

We meet such challenges by our innovative use of automation solutions, our sophisticated processes designed for operational efficiency, our highly skilled linguistic and IT leadership team and our commitment to quality.

Quality measures include transcriber recruitment and training, spell-checkers in our software tools and rigorous post-processing by senior team members in project and language specific issues. All transcribers are native speakers of the language.