Glossary
Code-switching (CS) is the practice of alternating between two or more languages within a single conversation, sentence, or even word. Bilingual and multilingual speakers do this constantly — Spanglish in the southern US, Hinglish in India, Singlish in Singapore. It's a major challenge for ASR systems, which traditionally assume a single language per recording.
Why code-switching is hard for ASR
Most ASR systems are language-tagged — you tell the model the audio is English, French, or Spanish, and it transcribes accordingly. Code-switching breaks this assumption. The model has to: (1) detect language at sub-utterance granularity, (2) switch its language model mid-sentence, and (3) produce correct orthography for each language. Older systems handled this terribly — typical strategy was to pick one dominant language and mis-transcribe the other.
How modern ASR handles it
Whisper was trained on multilingual audio and explicitly on code-switching examples. For common bilingual pairs (Spanish-English, Mandarin-English, Hindi-English), it produces reasonable transcripts with each segment in its source language. For rarer combinations (Tamil-Korean, Yoruba-French), accuracy drops sharply because the training data was sparse. Set Whisper's language parameter to auto for code-switched audio rather than tagging a single language.
Practical tips
If your audio is heavily code-switched: (1) don't tag a specific language — let the model auto-detect. (2) Expect ~5-10% lower accuracy than monolingual audio of the same quality. (3) Plan for a manual pass to clean up rare-language segments. (4) Consider speaker-by-speaker segmentation if speakers don't switch — diarize first, then run each speaker's segments with their dominant language tag.
In practice
A Spanglish family interview: "Then mi abuela said, 'Mijo, no vayas afuera, está raining.'" Whisper-3 transcribes this as a single utterance with each word in its source language and proper orthography. Older ASR systems (pre-2022) would have transcribed this as garbage — either pure English with phonetic Spanish gibberish, or vice versa.
Related terms
Further reading
Try multilingual transcription
Get started — freeLast updated: September 20, 2026
Frequently Asked Questions
What does Code-switching mean in transcription?
Mixing two or more languages within a single utterance — common in bilingual speakers. A challenging case for ASR; Whisper-3 handles common bilingual pairs (Spanish-English, Mandarin-English) but accuracy drops on rare combinations.
Why does Code-switching matter when choosing a transcription workflow?
Code-switching affects how you understand transcript quality, timing, compatibility, or the technology behind speech-to-text results.
Where can I apply Code-switching on Transcript.you?
Use the related workflow at /ai-transcription when you want to see how this glossary concept connects to an actual transcription task.
Browse the full transcription glossary or read the complete guide to AI transcription.