Fact · Speech recognition

Graves and colleagues wrote that recurrent networks needed pre-segmented training data and post-processing to produce label sequences, and presented a method that trains them to label unsegmented sequences directly.

Checked against its source on October 17, 2026

In the source's own words

“because they require pre-segmented training data, and post-processing to transform their outputs into label sequences, their applicability has so far been limited”

Connectionist Temporal Classification: Labelling Unsegmented Sequence Data with Recurrent Neural Networks, Alex Graves et al., Proceedings of the 23rd International Conference on Machine Learning (ICML 2006), pp. 369-376, 2006, Abstract.

Open the original (DOI 10.1145/1143844.1143891)

What it measures: The abstract names speech recognition as the motivating case: an acoustic signal transcribed into words or sub-word units.
What it does not show: This is the method later called connectionist temporal classification (CTC). Its experiments were on the small TIMIT corpus, not on large word-level tasks.

Source

Connectionist Temporal Classification: Labelling Unsegmented Sequence Data with Recurrent Neural NetworksPeer-reviewed research · Alex Graves et al., Proceedings of the 23rd International Conference on Machine Learning (ICML 2006), pp. 369-376, 2006

Graves and colleagues wrote that recurrent networks needed pre-segmented training data and post-processing to produce label sequences, and presented a method that trains them to label unsegmented sequences directly.

In the source's own words

Source

Cited on

Graves and colleagues wrote that recurrent networks needed pre-segmented training data and post-processing to produce label sequences, and presented a method that trains them to label unsegmented sequences directly.

In the source's own words

Source

Cited on