AI Voices for Poetry Narration

AI Voices for Poetry Narration

Poetry is the one text where the gaps carry as much as the words. These six come from the narration pool, chosen for an even delivery that leaves room for the line to land.

ServiceTitan
HVAC integration
HVAC integration
Trusted by 1,000+ users

Free to start!

Narrator ai voice
CormacVerified
British · Male · Young
United Kingdom flagEnglish
Use voice
AmeliaVerified
American · Female · Young
United States flagEnglish
Use voice
WesleyVerified
American · Male · Young
United States flagEnglish
Use voice
ReaganVerified
American · Female · Young
United States flagEnglish
Use voice
ManasiVerified
Indian · Female · Young
India flagHindi
Use voice
GopalVerified
Indian · Male · Young
India flagHindi
Use voice
CormacVerified
British · Male · Young
United Kingdom flagEnglish
Use voice
AmeliaVerified
American · Female · Young
United States flagEnglish
Use voice
WesleyVerified
American · Male · Young
United States flagEnglish
Use voice
ReaganVerified
American · Female · Young
United States flagEnglish
Use voice
ManasiVerified
Indian · Female · Young
India flagHindi
Use voice
GopalVerified
Indian · Male · Young
India flagHindi
Use voice

German AI Voices

Female Voices

Warm Voices

AI Voices for Podcast Generation

British English AI Voices

AI Voices for Documentary Voice Over

Calm Voices

Australian English AI Voices

AI Voices for Gaming & Character Voices

Professional Voices

AI Narrator Voices

Cheerful Voices

AI Voices for Customer Support Automation

Authoritative Voices

AI Voices for Meditation & Wellness

Gravelly AI Voices

AI Voices for YouTube Voiceover

Indian English AI Voices

AI Voices for Movie Trailers

Friendly Voices

Narration is the one job where a voice has to hold up for hours rather than seconds. 163 of the 244 voices here carry the narration tag, the largest group in the catalog, across English, Hindi and nine other Indic languages.

1500+ voices to explore

Browse our full voice library and find the right tone, accent, and personality for your agent.

Model Specifications

Streaming transports

WebSocket, HTTP chunked transfer

WebSocket, HTTP chunked transfer

Output formats

.ulaw, .alaw, PCM 16-bit, .wav, .mp3

.ulaw, .alaw, PCM 16-bit, .wav, .mp3

Speed range

0.5× – 2.0×, set per request

0.5× – 2.0×, set per request

Sample rate

44.1 kHz native, resampled to 8 kHz for telephony

44.1 kHz native, resampled to 8 kHz for telephony

Latency

~200 ms to first byte (p50, warm region)

~200 ms to first byte (p50, warm region)

Line breaks are instructions

Prose runs to punctuation, poetry runs to the line. A reading that ignores the break flattens the poem into a paragraph, which is the usual failure of machine-read verse. Because requests cap at 250 characters you generate line by line anyway, so each break becomes a real pause rather than something the model has to infer.

Pronunciation you can fix once

Poetry is full of archaisms, proper nouns and words that scan differently from how they are normally said. Pronunciation dictionaries let you define each one once so it holds across the whole collection. For anything metrical that matters more than voice choice, because a single mis-stressed word breaks the line.

Pacing carries the form

Speed adjusts from 0.5 to 2.0 and most verse sits between 0.8 and 0.9. A sonnet wants a different pace from free verse, and a slower read gives the stresses somewhere to land. That is the only delivery control available, since there is no tone or emotion parameter, so pace and line division are doing all the work.

Poetry in Indic languages

Hindi, Tamil, Telugu, Bengali and six more carry recommended voices, and 144 of 244 voices have an Indic recommendation. Recorded verse in regional Indian languages is thinly served everywhere, and the metrical traditions differ enough from English that a voice trained on the language matters more than usual.

Frequently asked questions

Generate line by line rather than passing the whole poem. Requests cap at 250 characters, which suits verse, and each separate call produces a real pause at the break instead of leaving the model to infer one.