Does it work in Igbo?

Results

Every feature gets one result. The result says what we can responsibly conclude, not what we suspect.

Supported

Current company documentation explicitly names Igbo for that specific feature, or a documented, repeatable test establishes it. Availability does not establish accuracy or dialect coverage.

Not supported in TokataHQ testing

TokataHQ, or community evidence that TokataHQ reviewed, tested the feature with spoken Igbo on the date shown, and Igbo was not available or did not work. This describes the version tested, not a company statement, and we retest as products change.

Company confirms no Igbo support

The company's current documentation explicitly excludes Igbo for that feature, including an expressly exhaustive language list. “We did not find Igbo listed” is never enough on its own.

Not yet tested

TokataHQ has not tested the feature, and company documentation does not settle it either way. This is neither a pass nor a failure.

Could not verify

There are early observations, but the record is incomplete, so we cannot yet make a reliable determination.

Evidence source

Separately from the result, each feature shows where its evidence came from.

Official documentation

The company's own current documentation, interface or published language list.

TokataHQ test

TokataHQ tested the feature directly. A complete record includes the date, device, app or model version, the Igbo sample used, the unedited output and the reviewer.

Community verified

A community member sent a screenshot, recording or description, and TokataHQ reviewed it. Unreviewed reports point us to what to test and never change a result on their own.

No test yet

Shown when there is no reliable evidence yet.

Dates

“Last tested” is when a hands-on test was run. “Last verified” is the date documentation or a complete test record established the result. A documentation review is not a product test.

Source order

1. Official company documentation.
2. Official product interfaces and published language lists.
3. TokataHQ hands-on testing.
4. Direct written company confirmation.
5. Community evidence reviewed by TokataHQ.
6. Reliable third-party technical documentation.

Search snippets, promotional claims, unsourced articles, and AI summaries do not establish definitive statuses. We read the underlying evidence and identify historical or inaccessible sources.

Keep the boundaries clear

Text translation, audio translation, transcription, captions, live translation, and speech generation are separate capabilities. An API is separate from a consumer app; a preview is separate from a general release. Model names and versions are recorded where known. Unknown versions or release stages remain explicitly unverified. Support for speech output (text-to-speech) never implies support for speech recognition, and the reverse.

A repeatable basic test

For every test, retain the product/model version, app or API surface, settings, device, date, exact inputs, unedited outputs, and reviewer. Record dialect, sample selection, number of attempts, and material failures. Share consented recordings or stable sample identifiers so another reviewer can repeat the test.

• Speech recognition (ASR), captions and transcription: Reuse a fixed set of short Igbo clips where practical. Have a fluent Igbo speaker prepare reference transcripts. State whether diacritics are expected. Separate intelligibility from exact transcription; retain errors and repetitions.
• Translation: Keep original text or audio and the complete output. Have an Igbo speaker review preservation of meaning and material errors. Test text-to-text and audio-to-text separately; timed subtitles require their own check.
• TTS: Use representative Igbo sentences, including names and tone-sensitive words. Have native speakers assess intelligibility and major pronunciation errors. Producing audio is not proof of high-quality speech.

Do not publish an accuracy percentage without a designed evaluation, stated sample size, reference answers, and scoring method. A small successful example cannot establish broad support, and one failed attempt describes only the version and setup tested. The three reported Gemini audio observations remain unverified pending a complete record.

Corrections and change

Technology changes quickly. Keep prior findings in the audit history, date the new evidence, and update only the capability it supports. Send a correction or a reproducible test record.