Norwegian Speech Recognition Acoustic Database from NST

A collection of acoustic data for Norwegian speech recognition and dictation, originally developed by Nordisk språkteknologi holding AS (NST_hesitate). The data was preserved after the company's 2003 bankruptcy and transferred to the National Library of Norway's Språkbanken in 2011. It is intended for developing automatic speech recognition systems.

Use Cases

• Train acoustic models for Norwegian automatic speech recognition using the provided audio recordings.
• Develop dictation systems for the Norwegian language based on the acoustic features captured in the database.
• Analyze and benchmark speech recognition performance on Norwegian acoustic data from the NST collection.
• Use the database to study phonetic and prosodic characteristics specific to Norwegian speech.

Strengths

• Data originates from a dedicated language technology firm (Nordisk språkteknologi holding AS) for speech recognition development.
• Preserved and made available through a national institution (National Library of Norway's Språkbanken) ensuring long-term access.
• The dataset was specifically created for the development of automatic speech recognition and dictation in Norwegian.

Limitations

• The original data collection ended with the company's bankruptcy in 2003, making the acoustic samples potentially outdated for modern speech patterns.
• The dataset's organization was altered post-acquisition, but the specific changes and their impact on usability are not detailed.
• Key technical specifications like audio format, sample rate, duration, and speaker demographics are unknown.

Provenance

Source

Nordisk språkteknologi holding AS (NST_hesitate), now held by the National Library of Norway's Språkbanken.

Collection Method

Developed by a commercial firm for speech recognition technology; acquisition details and original collection methodology are unknown.

Time Range

Developed prior to the company's bankruptcy in 2003.

Freshness

The metadata was last updated on 2024-09-10, but the underlying acoustic data is from pre-2003.

Geography

Norway (intended for the Norwegian language).