Voice / Single speaker

English text-to-speech

An original English narration made with Florence through the public voice studio: exact settings, measured time and the normal 8-credit charge are recorded.

Create with Donatello Voice Download original

Donatello VoiceEnglish text-to-speech

Created2026-10-13 02:13:05 UTC

Original size255 KB

Measured duration15.9s

Historical run charge8 credits

From input to reviewed output

Follow the workflow, inspect the evidence

A complete text-to-speech submission through the public production interface, followed by ordinary credit settlement and a downloadable MP3. The editorial account has an administrator role, but no administrative generation endpoint or billing bypass was used. Eight existing promotional credits were consumed.

Open the AI Voice Generator, choose Text to speech and paste the original 236-character English harbour narration recorded below. No reference recording or private cloned voice was used.
Select Florence from the English built-in catalogue. Keep normal speed, MP3, 128 kbps and normal latency. The recorded advanced values are temperature 0.7, top P 0.7, chunk length 300, volume 0 dB and both normalization options enabled.
Check the estimate before submitting. This run showed 8 credits, reserved them and consumed the same 8 on completion. It needed one attempt, used no new credit grant and created no PAYG reservation.
Open the completed result in History. Production timestamps measured 16.95 seconds from job acceptance to completion, including 0.431 seconds queued. The original MP3 measures about 15.88 seconds; typing, editorial review and download time are not part of that generation measurement.
Compare the complete file with the exact script. This record passed full decoding, local transcription and signal checks; it does not claim a subjective listening score. For your own publication, listen throughout and check pronunciation, pauses and rights before using the output.

Acceptance checks for this deliverable

The recorded Florence selection and exact narration match the production request.
The reservation and settlement agree: 8 credits, one successful attempt and no PAYG reservation.
The MP3 fully decodes and local transcription recovers all submitted words, allowing spelling and punctuation normalization.
Byte size, SHA-256, timing, signal levels and the untouched earlier QA audio are available in the run record.

What this record does not establish

This is not a non-admin-role test, new-account onboarding, Relax waiting-time test, cloning test or quality comparison. Automated transcription does not establish subjective naturalness. One observed generation time and historical charge do not guarantee future speed, price or pronunciation.

Voice safety and commercial-use guide →

Compare spoken-voice workflows and permissions →

Workflow notes updated 2026-10-13. The historical run, media, costs and creation date have not been changed.

Input record

Exact transcript

At sunrise, the harbour wakes slowly. A small boat crosses the water, and the bakery opens its blue door. For a moment, the whole town feels quiet and full of possibility. Today, we will follow one simple idea and see where it takes us.

Generation settings

Recorded controls

Public engine: Donatello Voice
Language: English
Selected voice: Florence · built-in catalogue voice
Output: MP3 · 44.1 kHz · mono
Voice mode: Text to speech
Delivery: 1× speed · 0 dB · normal latency · 128 kbps
Advanced controls: Temperature 0.7 · top P 0.7 · chunk length 300 · text and loudness normalization enabled
Generation attempts: 1
Timing definition: Production job creation to completion, including queue time; excludes typing, review and file download. One observation, not a latency guarantee.
Access path: Public production browser workflow with an editorial administrator account, ordinary queue and credit settlement. Eight existing promotional credits consumed; no billing bypass, new credit grant or PAYG charge. Recorded provider cost: $0.00354. Not a Relax or non-admin-role test.
Observed end-to-end time: 16.95s · one run, not a benchmark

Provenance

How this result was verified

Run verification

Matched to a completed production generation and its reservation/settlement ledger. The full MP3 decoded without errors; bytes, SHA-256, duration and audio levels were checked. Local speech recognition recovered the submitted words, allowing punctuation and harbour/harbor spelling differences. Automated first-party review, not a human listening panel.

Editing disclosure

No post-processing. The new production MP3 was copied unchanged. It replaces the featured QA sample; the earlier file and its historical record remain available below.

Interpretation limit

One English catalogue-voice run on an editorial administrator account. No administrative generation endpoint or billing bypass was used; this is not a non-admin-role, registration, Relax, cloning, dialogue or multilingual test. No quality winner or guaranteed pronunciation is established.

SHA-256 of the public original83d9341d1e2731827ea1ac8984d29c04cf9876250ec2435dc374af8b16d0b93a254,998 bytes · audio/mpeg

Rights and reuse

Commercial-use policy, with the important conditions intact.

Donatello permits users to use their generated outputs commercially under the Terms, subject to lawful inputs, third-party rights and applicable law. This first-party sample is published as evidence and may be cited with attribution; do not treat the gallery file itself as stock media. Generate your own output for a commercial project.