Emolia-HQ: High-Quality Speaker-Paired Audio for Voice Synthesis

A high-quality, speaker-paired subset of the LAION Emolia dataset, created by TTS-AGI and last updated on March 9,我们发现 2026. Each sample includes a target and a reference utterance from the same speaker, filtered for quality using a DNSMOS score threshold of 3.0.

Use Cases

• Voice conversion based on speaker-paired target and reference utterances.
• Expressive text-to-speech synthesis conditioned on speaker identity.
• Speaker-aware emotion recognition using paired audio samples.

Strengths

• Quality-filtered using a DNSMOS score threshold of 3.0.
• Each sample is paired with a reference audio from the same speaker, enabling speaker-conditioned tasks.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.

Provenance

Source

Derived from the laion/Emolia dataset.

Collection Method

Created through quality filtering and speaker pairing.

Time Range

null

Freshness

Last updated 2026-03-09 06:59:16; freshness should be verified.

Geography

null