Contents Menu Expand Light mode Dark mode Auto light/dark mode
S3PRL 0.4.17 documentation
S3PRL 0.4.17 documentation

Getting started

  • Install S3PRL
  • S3PRL Upstream Collection
  • Use Problem module to run customizable recipes

How to Contribute

  • General Guideline
  • Adding New Upstream

API Documentation

  • nn
    • beam_decoder
    • common
    • hear
    • interface
    • linear
    • pit
    • pooling
    • rnn
    • speaker_loss
    • speaker_model
    • specaug
    • upstream
  • problem
    • asr
      • run
      • superb_asr
      • superb_pr
      • superb_sf
    • asv
      • run
      • superb_asv
    • base
    • common
      • example
      • hear_beijing_opera
      • hear_cremad
      • hear_dcase_2016_task2
      • hear_esc50
      • hear_fsd
      • hear_gsc5hr
      • hear_gtzan
      • hear_gtzan_music_speech
      • hear_gunshot
      • hear_libricount
      • hear_maestro
      • hear_nsynth5hr
      • hear_stroke
      • hear_tonic
      • hear_vocal
      • hear_vox_lingual
      • run
      • superb_er
      • superb_ic
      • superb_ks
      • superb_sid
    • diarization
      • run
      • superb_sd
      • util
  • task
    • base
    • diarization
    • dump_feature
    • event_prediction
    • scene_prediction
    • speaker_verification_task
    • speech2text_ctc_task
    • utterance_classification_task
  • dataio
    • collate_fn
    • corpus
      • base
      • fluent_speech_commands
      • iemocap
      • librilight
      • librispeech
      • quesst14
      • snips
      • speech_commands
      • voxceleb1sid
      • voxceleb1sv
    • dataset
      • base
      • diarization
      • encode
      • frame_label
      • load_audio
      • util
    • encoder
      • category
      • g2p
      • tokenizer
      • vocabulary
    • sampler
      • balanced_weighted_sampler
      • distributed_sampler
      • fixed_batch_size_batch_sampler
      • group_same_item_sampler
      • max_timestamp_batch_sampler
      • sorted_sampler
  • metric
    • common
    • diarization
    • slot_filling
  • util
    • audio_info
    • benchmark
    • download
    • override
    • pseudo_data
    • seed
Back to top

dataio#

(s3prl.dataio)

This package handles data-related sub-tasks

s3prl.dataio.collate_fn

s3prl.dataio.corpus

Parse the commonly used corpus into standardized dictionary structure

s3prl.dataio.dataset

s3prl.dataio.encoder

Encode the raw data into numeric format, and then decode it

s3prl.dataio.sampler

Control how torch DataLoader group instances into a batch

Next
collate_fn
Previous
utterance_classification_task
Copyright © 2022, S3PRL Team
Made with Sphinx and @pradyunsg's Furo