Multilingual voices
Multilingual AI voices: one voice, many languages
SpeakScribe has 12 multilingual voices that read text in other languages with the same voice — for example, Andrew Multilingual can narrate an English video and its Spanish version. We tested every one of them in 8 languages: Whisper recognised the language correctly in 96 of 96 clips.
The 12 multilingual voices
Each voice has a home language and accent and reads other languages with the same timbre. Play each one in its home language, in Spanish and in Hindi.
Female · home language American English
en-US-AvaMultilingualNeural
Use in Voice Studio →
Female · home language American English
en-US-EmmaMultilingualNeural
Use in Voice Studio →
Male · home language American English
en-US-AndrewMultilingualNeural
Use in Voice Studio →
Male · home language American English
en-US-BrianMultilingualNeural
Use in Voice Studio →
Male · home language Australian English
en-AU-WilliamMultilingualNeural
Use in Voice Studio →
Male · home language Korean
ko-KR-HyunsuMultilingualNeural
Use in Voice Studio →
Female · home language French
fr-FR-VivienneMultilingualNeural
Use in Voice Studio →
Male · home language French
fr-FR-RemyMultilingualNeural
Use in Voice Studio →
Female · home language German
de-DE-SeraphinaMultilingualNeural
Use in Voice Studio →
Male · home language German
de-DE-FlorianMultilingualNeural
Use in Voice Studio →
Female · home language Brazilian Portuguese
pt-BR-ThalitaMultilingualNeural
Use in Voice Studio →
Male · home language Italian
it-IT-GiuseppeMultilingualNeural
Use in Voice Studio →
One male voice for English and Spanish
If you publish the same video in English and Spanish and want one narrator for both, use a male multilingual voice:
- → Andrew Multilingual — home language American English
- → Brian Multilingual — home language American English
- → William Multilingual — home language Australian English
- → Hyunsu Multilingual — home language Korean
- → Remy Multilingual — home language French
- → Florian Multilingual — home language German
- → Giuseppe Multilingual — home language Italian
Female multilingual voices: Ava, Emma, Vivienne, Seraphina, Thalita.
When a native voice is the better choice
A multilingual voice keeps one character across languages. When the audience should hear a local accent, a native voice fits better:
- US / Latin American Spanish: Alonso, Jorge (Mexico)
- Spain: Alvaro
- Every language: browse all 75 languages
We tested every multilingual voice in 8 languages
On 2026-10-16 each voice read the same two sentences in 8 languages. We then ran every clip through Whisper large-v3-turbo with automatic language detection and compared the transcript with the original text. A ✓ means Whisper identified the intended language; the number is how closely its transcript matched the original (100% = identical). This measures whether the speech is clearly understood as that language — not how natural it sounds, which is best judged by listening.
| Voice | English | Spanish | French | German | Portuguese | Hindi | Arabic | Japanese |
|---|---|---|---|---|---|---|---|---|
| Ava (English) | ✓ 100% | ✓ 100% | ✓ 100% | ✓ 100% | ✓ 100% | ✓ 95% | ✓ 98% | ✓ 97% |
| Emma (English) | ✓ 100% | ✓ 100% | ✓ 100% | ✓ 100% | ✓ 100% | ✓ 100% | ✓ 98% | ✓ 95% |
| Andrew (English) | ✓ 100% | ✓ 100% | ✓ 100% | ✓ 100% | ✓ 95% | ✓ 98% | ✓ 94% | ✓ 97% |
| Brian (English) | ✓ 100% | ✓ 100% | ✓ 100% | ✓ 100% | ✓ 96% | ✓ 100% | ✓ 98% | ✓ 97% |
| William (English) | ✓ 100% | ✓ 100% | ✓ 100% | ✓ 100% | ✓ 96% | ✓ 98% | ✓ 98% | ✓ 95% |
| Hyunsu (Korean) | ✓ 100% | ✓ 100% | ✓ 100% | ✓ 100% | ✓ 100% | ✓ 98% | ✓ 98% | ✓ 97% |
| Vivienne (French) | ✓ 100% | ✓ 100% | ✓ 100% | ✓ 100% | ✓ 96% | ✓ 100% | ✓ 98% | ✓ 97% |
| Remy (French) | ✓ 95% | ✓ 100% | ✓ 100% | ✓ 100% | ✓ 96% | ✓ 98% | ✓ 98% | ✓ 97% |
| Seraphina (German) | ✓ 95% | ✓ 100% | ✓ 100% | ✓ 100% | ✓ 96% | ✓ 98% | ✓ 98% | ✓ 95% |
| Florian (German) | ✓ 95% | ✓ 97% | ✓ 100% | ✓ 100% | ✓ 96% | ✓ 100% | ✓ 98% | ✓ 97% |
| Thalita (Portuguese) | ✓ 100% | ✓ 100% | ✓ 100% | ✓ 100% | ✓ 100% | ✓ 98% | ✓ 98% | ✓ 95% |
| Giuseppe (Italian) | ✓ 100% | ✓ 100% | ✓ 100% | ✓ 100% | ✓ 100% | ✓ 100% | ✓ 98% | ✓ 95% |
Matches below 100% are mostly formatting differences, such as Whisper writing “3” for “three”.
For comparison: standard voices reading another language
We ran the same test on three voices that are not multilingual:
- → Jenny (American English) reading Spanish: recognised as Spanish, 100% match.
- → Jenny (American English) reading Hindi: no audio was produced.
- → Jenny (American English) reading French: heard by Whisper as English instead of French (100% match).
- → Jorge (Mexican Spanish) reading English: recognised as English, 95% match.
- → Jorge (Mexican Spanish) reading French: heard by Whisper as Spanish instead of French (73% match).
- → Andrew (American English) reading Spanish: heard by Whisper as English instead of Spanish (100% match).
- → Andrew (American English) reading Hindi: heard by Whisper as Urdu instead of Hindi (24% match).
Standard voices can sometimes read another language, but often with an accent strong enough to be mistaken for their own language, and some scripts fail completely. That is why the multilingual voices exist.
How to use a multilingual voice
- Open Voice Studio and choose a multilingual voice.
- Paste the script in the target language — the voice follows the language of the text; there is no language setting.
- For bilingual scripts, keep each language in its own paragraph so the switch falls on a pause.
- Write the script in the language's own script (Devanagari for Hindi, Arabic script for Arabic) — romanised text is read with the voice's home pronunciation.
- Generate and download the MP3. Translating a video? See the multilingual workflow.
Questions about multilingual voices
Make one video in two languages
Pick a multilingual voice, paste each version of your script, and download both voiceovers.
Open Voice Studio