ViVoice-34: Vietnamese Speech Recordings from 34 Provinces

ViVoice-34 is a Vietnamese speech dataset featuring audio recordings from speakers across 34 provinces of Vietnam. Each audio sample includes full transcripts and metadata about the speaker and content. The dataset was created by anonymous-vivoice34 and was last updated on Hugging Face in May 2026.

Use Cases

• Train Vietnamese automatic speech recognition models based on the provided audio transcripts.
• Develop speaker identification or verification systems based on the metadata about speakers from different provinces.
• Fine-tune text-to-speech models for Vietnamese using the paired audio and text data.
• Analyze regional accents or dialects in Vietnamese speech based on the geographic speaker metadata.

Strengths

• Includes speech samples from speakers across 34 different provinces, suggesting geographic diversity.
• Each audio sample is paired with a full transcript, enabling supervised learning tasks.

Limitations

• The audio samples in the main repository are sample-only; the full dataset must be downloaded separately, which complicates immediate inspection.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.

Provenance

Source

anonymous-vivoice34 on Hugging Face

Collection Method

Likely contains recordings collected from speakers across Vietnam.

Freshness

Last updated 2026-05-08 09:59:52; freshness should be verified.

Geography

Vietnam, covering 34 provinces.