German Speech Recordings From Mobile Phone Readings

Encompassing a sample of 327 German native speakers reading texts into mobile phones. The recordings cover economics, entertainment, news, and other topics, with each sentence averaging 10.3 words and being repeated 1.4 times on average. All audio has been manually transcribed.

Use Cases

• Train an automatic speech recognition model on manually transcribed German speech from mobile phone recordings.
• Analyze phonetic or prosodic patterns in German speech using the 1.4 average repetitions per sentence.
• Develop a text-to-speech model for German using the pairing of high-accuracy transcriptions with native speaker audio.
• Study speech variation across different reading content types like economics, entertainment, and news.

Strengths

• Includes 327 unique German native speaker recordings.
• All transcriptions are manually created to ensure high accuracy.
• Sentences average 10.3 words, providing substantial linguistic content per audio sample.
• Covers diverse reading content including economics, entertainment, and news.

Limitations

• The dataset is described as a sample, implying it is not the full collection.
• The specific audio format, sample rate, and total duration are unknown.
• Limited demographic or geographic information about the 327 speakers is provided.

Provenance

Source

Nexdata

Collection Method

Recordings of German native speakers reading texts into mobile phones.

Freshness

The dataset card was last updated on 2025-04-23.