ITR Atlas
← All interfaces

Speech Recognition, composition (Foley et al., 2020)

Measured by Foley, Casiez & Vogel · ACM CHI 2020 (2020)

Voice Speech recognition 2020

Inputs

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

rate = 117 wpm
Speech rate composing original text by voice on a Google Pixel 3 with the default recognizer; touchscreen typing in the same composition task was 35 wpm. Composition is slower than read-aloud transcription because the user is also deciding what to say.
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
Modern ASR, composition task
9.75 bits/s

  1. Characters per minute

    117 wpm × 5 chars/word = 585 chars/min

    Residual uncorrected error of the final composed text was ~0.5–0.65% (speech had a lower error rate than touchscreen typing). The 117 wpm is net of correction time, so the error is documented here, not re-applied.

  2. Bits per character

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

  3. Information transfer rate

    585 char/min × 1.0 bit/char ÷ 60 s/min = 9.75 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
Open-vocabulary ASR with a language model in the loop, scored on composed (not read-aloud) text. The bits/s figure uses an English character-entropy proxy; the per-utterance action space is effectively unbounded, so this is not a fixed-target selection.

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
Foley, Casiez & Vogel
Publication
ACM CHI 2020, 2020
Paper
10.1145/3313831.3376861