ITR Atlas
← All interfaces

Speech Recognition, early ASR (Karat et al., 1999)

Measured by Karat, Halverson, Horn & Karat · ACM CHI 1999 (1999)

Voice Speech recognition 1999

Inputs

The measured or assumed values behind the calculations, each with its source.

rate = 25 wpm
Effective transcription rate AFTER error correction for experienced users on late-1990s large-vocabulary continuous dictation; raw dictation was ~107 wpm before correction. New users averaged ~14 wpm transcribing and ~8 wpm composing.
H = 1.0 bits/char
English-text entropy (Shannon).

Strictest ITR

Each scoring method is an upper bound on the channel, so the headline is the strictest (smallest) one for this entry. Use the score selector on the home page to view any single method across entries.

Strictest Shannon (text) Recomputed
Character-entropy throughput
Early ASR, corrected transcription
2.08 bits/s

  1. Characters per minute

    25 wpm (corrected) × 5 chars/word = 125 chars/min

    Recognition errors were frequent: raw dictation ran ~105–107 wpm but correction cut the effective rate to ~25 wpm. The error shows up here as that correction overhead rather than a published WER%, and it is already in the corrected rate, not re-applied.

  2. Bits per character

    H(English) ≈ 1.0 bit/char (Shannon)

  3. Information transfer rate

    125 char/min × 1.0 bit/char ÷ 60 s/min = 2.08 bits/s

What counts as a bit depends on the action space. The number of distinguishable actions and how likely each one is are design choices of the task, not the sensing hardware. The same modality can present a fixed set of targets, a set pruned per step by a grammar or language model, or a continuous control space. Each of these changes how many actions are live and how the probability mass is spread, and therefore the information per selection. Read the action space below before comparing headline numbers across entries.

Action space

What the user can produce at each step, and how those options are distributed.

Structure
Context-dependent (the live set changes per step)
Size
Continuous
Prior
Context-conditioned: likelihoods depend on prior actions
Notes
Late-1990s desktop ASR (IBM ViaVoice / Dragon era). Recognition errors were frequent, so correction time dominated: ~107 wpm raw dictation collapsed to ~25 wpm effective. This is the historical floor that modern deep-learning ASR (see the 2016 transcription and 2020 composition speech entries) later lifted ~6×.

Comparability The strictest bound here is the Shannon entropy of the output text, under one predictor held constant across the whole atlas (≈1 bit per character). That shared predictor makes it directly comparable to every other text entry (keyboards, spellers, silent speech and speech BCIs) regardless of prior or vocabulary size. For most text interfaces it comes out tighter than the raw-selection bounds, but not always. Where a small vocabulary makes Wolpaw tighter, that wins instead. Any Fitts, Wolpaw or log₂(N) figure shown below is another bound on the same channel. Switch the home-page score selector to compare one across entries.

Source

Authors
Karat, Halverson, Horn & Karat
Publication
ACM CHI 1999, 1999
Paper
10.1145/302979.303160