Adult Voice: Mandarin TTS Average Voice Library with Balanced Age and Gender

A Mandarin-accented average voice library for text-to-speech, released in October 2024. It contains recordings from approximately 200 participants, balanced between males and females aged 18-70, with 100 speakers from southern China and 100 from northern China. The data was created by longmaodata and features quiet recording environments with a signal-to-noise ratio greater than 20 dB.

Use Cases

• Training average voice models for Mandarin TTS based on the balanced demographic data.
• Developing multi-speaker TTS systems based on the 200-speaker corpus.
• Researching accent and regional variation in Mandarin speech based on the north-south speaker split.
• Building speech synthesis systems requiring high-quality, clean audio based on the SNR>20 dB specification.

Strengths

• Demographic balance across age (18-70) and gender (male/female).
• Approximately 200 participants, split evenly between northern and southern China.
• High-quality recording conditions with a signal-to-noise ratio greater than 20 dB.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count, file formats, and license information are unknown, which may limit suitability assessment.

Provenance

Source

longmaodata via Hugging Face

Collection Method

Likely contains recorded speech from participants under controlled conditions.

Freshness

Last updated 2024-11-22 06:39:11; freshness should be verified.

Geography

China (speakers from northern and southern regions)