Mozilla Common Voice
An open-source voice dataset project initiated by Mozilla, containing multilingual speech samples. It is commonly used for training and evaluating speech synthesis and recognition models.
Timeline (last 90 days)
According to independent benchmarks such as Mozilla Common Voice MOS, ElevenLabs consistently scores 4.6-4.8, while Azure usually ranges between 4.2-4.4.
CoVoST2 is a large-scale speech translation dataset built on Mozilla Common Voice, containing translation directions for multiple language pairs.
All Facts (2)
According to independent benchmarks such as Mozilla Common Voice MOS, ElevenLabs consistently scores 4.6-4.8, while Azure usually ranges between 4.2-4.4.
50%UnverifiedCoVoST2 is a large-scale speech translation dataset built on Mozilla Common Voice, containing translation directions for multiple language pairs.
50%