Librispeech Alignments: 1000 Hours of English Speech with Forced Alignments

Librispeech is a 1000-hour corpus of 16kHz read English speech derived from audiobooks, designed for automatic speech recognition. This version includes alignments generated by the Montreal Forced Aligner (MFA). The dataset was uploaded to Hugging Face by gilkeyio and last updated on November 22, 2023.

Use Cases

• Training ASR models based on the 1000 hours of labeled English speech.
• Evaluating forced alignment algorithms using the provided MFA-generated alignments.
• Developing speech synthesis or prosody models based on the alignment of phonemes to audio.
• Researching read speech patterns and pronunciation using the audiobook-derived corpus.

Strengths

• Contains 1000 hours of audio data, providing substantial scale for model training.
• Includes forced alignments generated by a standard tool (Montreal Forced Aligner), adding structural metadata.
• Derived from a well-known benchmark corpus (Librispeech) for speech recognition.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and file formats are unknown, which may limit suitability assessment.
• Data may reflect bias inherent to the source audiobooks, such as genre or narrator demographics.

Provenance

Source

Librispeech corpus, processed by gilkeyio.

Collection Method

Alignments generated by the Montreal Forced Aligner (MFA).

Time Range

null

Freshness

Last updated 2023-11-22 16:03:34; freshness should be verified.

Geography

null