Edge TTS voices list: 322 voices and the best for narration
Microsoft offers hundreds of neural voices through Edge TTS. Here is the list in numbers, what the tags mean, and a short test for picking one narrator.
The Edge TTS voices list is the set of Microsoft neural voices available through the online service behind Read Aloud in Microsoft Edge. The open-source edge-tts tool prints the list with one command, and on 2026-11-01 the list held 322 voices across 142 locales: 163 female and 159 male. Each voice has a name, a locale such as en-US or vi-VN, a gender, a category and a personality tag. Most voices, 298 of 322, carry the "General" category, while a few carry "News, Novel" or "Conversation, Copilot". For example, en-US-GuyNeural is tagged "News, Novel" and "Passion", which hints at a narrator voice. The guide first explains the tags, then the English and Vietnamese options. Finally, the guide shows a short test for picking one narrator and how to use the chosen voice for batch MP3 export with Tofu Voice from TofuToon.
What do the Edge TTS voices list tags mean?
The tags in the Edge TTS voices list are short labels from Microsoft that describe where a voice fits and how it sounds. The category names the intended use; the personality tag names the tone. In the list printed on 2026-11-01, the categories were:
- General: 298 voices, the default for most locales.
- News, Novel: 8 voices, suited to narration and reading.
- Conversation, Copilot: 8 voices, tuned for chat-style speech.
- Cartoon, Novel and Cartoon, Conversation: 3 voices for character work.
- Dialect: 2 voices for regional speech.
Personality tags include "Warm", "Rational", "Friendly" and "Authority". In practice, a "News, Novel" voice with a "Reliable, Authority" tag is a natural first test for audiobooks, while "Cheerful, Clear" fits short explainers. Tags are hints from Microsoft, not promises, so the real test is your own script read aloud.
Which English Edge TTS voices suit narration?
The en-US locale is the main source of English narration voices in the Edge list, with 17 voices on 2026-11-01, plus British, Indian and other English locales. The "News, Novel" group includes Guy ("Passion"), Christopher ("Reliable, Authority"), Eric and Steffan ("Rational"), Roger ("Lively") and Aria ("Positive, Confident"). The "Conversation, Copilot" group includes Andrew ("Warm, Confident"), Brian ("Approachable, Casual"), Ava and Emma. Twelve voices in the list are multilingual versions, such as AndrewMultilingual, which can read more than one language. First, shortlist two male and two female voices from the "News, Novel" group for a story or audiobook. Second, test the shortlist on the same paragraph. For instance, Christopher and Aria give a clear contrast for a first test. British voices such as Ryan, Sonia and Thomas, tagged "Friendly, Positive" by Microsoft, suit a UK audience.
Which Vietnamese Edge TTS voices are there?
Vietnamese has two voices in the Edge TTS list: vi-VN-HoaiMyNeural, female, and vi-VN-NamMinhNeural, male, both tagged "General" and "Friendly, Positive" on 2026-11-01. Two voices is a short list, but both read tone marks natively, which matters more than variety for long Vietnamese narration. First, test both voices on a paragraph with hỏi and ngã marks, a name and a long number. Second, pick the voice that suits the content: HoaiMy for a warm narrator, NamMinh for a calm male narrator. Finally, keep the voice for the whole series. For more Vietnamese options, CapCut lists 24 Vietnamese voices in the catalog bundled with Tofu Image Text (counted on 2026-10-31), including news and review styles. In practice, a Vietnamese audiobook channel can test HoaiMy, NamMinh and two CapCut voices in one session using Tofu Voice, which offers both groups.
How do you pick one narrator from the Edge TTS voices list?
Picking a narrator from 322 voices is a filter, then a listening test. First, filter by locale: only native voices for the language of the script. Second, filter by category: "News, Novel" for stories and audiobooks, "General" for most other content. Third, export the same paragraph with three or four candidates. Finally, listen on phone speakers and headphones, then for five minutes in a row. Tofu Voice is a pay-once desktop app for Windows 10 or later and Apple silicon Macs (M1 or later) that offers Edge and CapCut voices in Base, with 170+ natural voices in total, according to TofuToon. For instance, a test paragraph exported with four voices gives four short MP3 files to compare. Once chosen, the same voice converts a whole folder of chapter files, one MP3 per file. Edge voices are online, so the text goes to Microsoft while generating.
FAQ
How do I see the full Edge TTS voices list myself?
The full Edge TTS voices list is one command away with the open-source edge-tts tool: install the package with Python, then run edge-tts with the --list-voices option. The output prints every voice with its name, gender, category and personality tag. Microsoft also publishes a language and voice support page for Azure AI Speech, the paid developer service, which lists neural voices by locale. The list printed on 2026-11-01, for example, held 322 voices; a later run can show more or fewer as Microsoft changes the catalog. First, save the output to a text file with the date. Second, search the file for your locale, such as vi-VN or en-US. Finally, test the short list by ear. Tofu Voice shows its own voice list in the app, with a preview button for each single voice.
Are Edge TTS voices good for audiobooks?
Edge TTS voices are a good fit for audiobooks when the voice is a calm native speaker and the text is clean. The "News, Novel" category from Microsoft exists for reading, and voices such as Christopher, Guy and Aria are natural first tests in English. For Vietnamese, HoaiMy and NamMinh read tone marks natively. First, clean the text: remove page numbers, headers and symbols. Second, test the voice on a full page, not a sentence. Finally, listen for five minutes in a row, since fatigue only shows over time. For example, a 300-page novel at 150 words per minute runs about 10 hours, so a voice that grates after ten minutes will lose listeners. Check the terms from Microsoft before commercial use, since the voices come from a Microsoft online service with its own rules.
What are multilingual Edge TTS voices?
Multilingual Edge TTS voices are voices whose names include "Multilingual", such as AndrewMultilingual, AvaMultilingual and EmmaMultilingual, and the list printed on 2026-11-01 held 12 of them. A multilingual voice can read text in more than one language with the same voice identity, which helps a bilingual channel keep one narrator. Native single-language voices can still sound more natural, especially for tonal languages such as Vietnamese. For example, an English and Spanish channel can test AvaMultilingual on both scripts and compare the Spanish result with a native es-ES voice. First, test the multilingual voice on each language. Second, compare each result with a native voice. Finally, keep whichever sounds right in each language, even with two narrators.
Next step: shortlist three voices from the "News, Novel" group or your native locale, export one test paragraph with each, and keep the clearest. For batch MP3 export with Edge or CapCut voices, see Tofu Voice, and read Edge TTS without code for the tools.