Fact · Languages and accents
42data points
The on-device speech recognition across English accents (EdAcc conversations) dataset holds 42 data points, each a published figure traced to its source.
Calculated from 42 data points on October 21, 2026
- How it is calculated
- Count of the published facts that are rows of the dataset: 42.
A statistic of the dataset Speech Recognition and Accents: 13 English Accents Tested, calculated by code from its published rows.
Calculated from
- In our October 2026 test on a Mac with an M2 Pro, Apple SpeechTranscriber had a word error rate of 16.39 percent on 1224 conversational utterances by speakers of all 13 accent groups together (47 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Apple SpeechTranscriber had a word error rate of 14.11 percent on 83 conversational utterances by Chinese speakers (2 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Apple SpeechTranscriber had a word error rate of 13.29 percent on 100 conversational utterances by speakers of Indian English (3 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Apple SpeechTranscriber had a word error rate of 11.49 percent on 99 conversational utterances by speakers of Irish English (3 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Apple SpeechTranscriber had a word error rate of 19.49 percent on 100 conversational utterances by speakers of Jamaican English (4 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Apple SpeechTranscriber had a word error rate of 17.29 percent on 100 conversational utterances by speakers of Kenyan English (2 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Apple SpeechTranscriber had a word error rate of 10.77 percent on 52 conversational utterances by Latin American Spanish speakers (2 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Apple SpeechTranscriber had a word error rate of 13.81 percent on 99 conversational utterances by Lithuanian speakers (2 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Apple SpeechTranscriber had a word error rate of 15.45 percent on 97 conversational utterances by speakers of Mainstream US English (9 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Apple SpeechTranscriber had a word error rate of 20.20 percent on 99 conversational utterances by speakers of Nigerian English (5 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Apple SpeechTranscriber had a word error rate of 15.65 percent on 98 conversational utterances by speakers of Scottish English (2 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Apple SpeechTranscriber had a word error rate of 17.91 percent on 98 conversational utterances by speakers of Southern British English (2 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Apple SpeechTranscriber had a word error rate of 17.01 percent on 99 conversational utterances by Spanish speakers (5 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Apple SpeechTranscriber had a word error rate of 25.71 percent on 100 conversational utterances by Vietnamese speakers (6 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Parakeet TDT 0.6B v3 had a word error rate of 12.57 percent on 1224 conversational utterances by speakers of all 13 accent groups together (47 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Parakeet TDT 0.6B v3 had a word error rate of 9.70 percent on 83 conversational utterances by Chinese speakers (2 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Parakeet TDT 0.6B v3 had a word error rate of 9.82 percent on 100 conversational utterances by speakers of Indian English (3 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Parakeet TDT 0.6B v3 had a word error rate of 9.35 percent on 99 conversational utterances by speakers of Irish English (3 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Parakeet TDT 0.6B v3 had a word error rate of 16.01 percent on 100 conversational utterances by speakers of Jamaican English (4 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Parakeet TDT 0.6B v3 had a word error rate of 12.17 percent on 100 conversational utterances by speakers of Kenyan English (2 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Parakeet TDT 0.6B v3 had a word error rate of 8.31 percent on 52 conversational utterances by Latin American Spanish speakers (2 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Parakeet TDT 0.6B v3 had a word error rate of 10.54 percent on 99 conversational utterances by Lithuanian speakers (2 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Parakeet TDT 0.6B v3 had a word error rate of 12.32 percent on 97 conversational utterances by speakers of Mainstream US English (9 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Parakeet TDT 0.6B v3 had a word error rate of 13.44 percent on 99 conversational utterances by speakers of Nigerian English (5 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Parakeet TDT 0.6B v3 had a word error rate of 13.11 percent on 98 conversational utterances by speakers of Scottish English (2 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Parakeet TDT 0.6B v3 had a word error rate of 11.83 percent on 98 conversational utterances by speakers of Southern British English (2 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Parakeet TDT 0.6B v3 had a word error rate of 13.93 percent on 99 conversational utterances by Spanish speakers (5 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Parakeet TDT 0.6B v3 had a word error rate of 22.76 percent on 100 conversational utterances by Vietnamese speakers (6 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Whisper large-v3-turbo had a word error rate of 17.66 percent on 1224 conversational utterances by speakers of all 13 accent groups together (47 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Whisper large-v3-turbo had a word error rate of 13.50 percent on 83 conversational utterances by Chinese speakers (2 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Whisper large-v3-turbo had a word error rate of 14.26 percent on 100 conversational utterances by speakers of Indian English (3 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Whisper large-v3-turbo had a word error rate of 11.24 percent on 99 conversational utterances by speakers of Irish English (3 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Whisper large-v3-turbo had a word error rate of 20.44 percent on 100 conversational utterances by speakers of Jamaican English (4 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Whisper large-v3-turbo had a word error rate of 19.21 percent on 100 conversational utterances by speakers of Kenyan English (2 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Whisper large-v3-turbo had a word error rate of 12.07 percent on 52 conversational utterances by Latin American Spanish speakers (2 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Whisper large-v3-turbo had a word error rate of 19.33 percent on 99 conversational utterances by Lithuanian speakers (2 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Whisper large-v3-turbo had a word error rate of 16.93 percent on 97 conversational utterances by speakers of Mainstream US English (9 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Whisper large-v3-turbo had a word error rate of 22.03 percent on 99 conversational utterances by speakers of Nigerian English (5 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Whisper large-v3-turbo had a word error rate of 15.49 percent on 98 conversational utterances by speakers of Scottish English (2 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Whisper large-v3-turbo had a word error rate of 17.53 percent on 98 conversational utterances by speakers of Southern British English (2 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Whisper large-v3-turbo had a word error rate of 18.82 percent on 99 conversational utterances by Spanish speakers (5 speakers) from the EdAcc corpus.
- In our October 2026 test on a Mac with an M2 Pro, Whisper large-v3-turbo had a word error rate of 28.21 percent on 100 conversational utterances by Vietnamese speakers (6 speakers) from the EdAcc corpus.