Part 1. Overview

FineVoice Text to Speech (TTS) delivers ultra-realistic, context-aware voice synthesis and precise voice cloning. It supports multiple voice models and emotion tags, enabling users to generate high-quality voice content for various scenarios—including learning, dubbing, content creation, and business services.

Prerequisites

Before getting started, please prepare the following:

A FineVoice account
Stable internet connection
The latest version of Google Chrome or Edge
For the best experience, use on a computer

TTS Models

FineVoice TTS

FineVoice TTS Max

The model delivers high-quality, low-latency voice generation.

Realistic voices with low latency
Supports voice styles and effects
Supports 154 languages
Supports voice cloning models
Accessible via API

Our most powerful and expressive TTS model, featuring emotion tags and the ability to mimic natural speech rhythm and emotion.

More expressive, context-aware speech
Enhanced multilingual capabilities
Supports emotion tags
Supports voice cloning models
Supports voice design models
Accessible via API

Limitations and Benefits

	Free	Basic	Pro	Enterprise
Input Characters	250	5,000	5,000	5,000
Result Download	No	Yes	Yes	Yes
Conversion Quota	2,000 TTS characters/month	100,000 TTS characters/month	300,000 TTS characters/month	1,000,000 TTS characters/month

Check Full FineVoice Pricing >>

AI Voices

FineVoice offers a diverse range of built-in voices across languages, professions, and ages, plus a community-generated voice model library. The FineVoice voice library showcases rich vocal characteristics and styles, all available for trial in FineVoice Text to Speech.

For custom needs, use our Voice Cloning and Voice Design tools to easily create the perfect voice for your project.

Currently, FineVoice has discontinued public access to community cloning models. Use of cloning models must comply with ethical and legal standards.

For best results, ensure the input text matches the original voice's language. While cross-language generation is possible, the output will retain the original voice model's accent, and the quality, pronunciation, and naturalness may be reduced.

Language Support

FineVoice TTS supports 154 languages.
FineVoice TTS Max supports the following languages and regions:

zh-CN — Chinese (Mainland)
zh-HK — Chinese (Hong Kong)
en-US — English (United States)
en-GB — English (United Kingdom)
en-AU — English (Australia)
en-HK — English (Hong Kong)
en-ZA — English (South Africa)
en-TZ — English (Tanzania)
en-SG — English (Singapore)
en-PH — English (Philippines)
en-NZ — English (New Zealand)
en-NG — English (Nigeria)
en-KE — English (Kenya)
en-IN — English (India)
en-IE — English (Ireland)
de-DE — German (Germany)
fr-FR — French (France)
es-ES — Spanish (Spain)
ar-AR — Arabic (General/Region)
ko-KR — Korean (South Korea)
ru-RU — Russian (Russia)
nl-BE — Dutch (Belgium)
nl-NL — Dutch (Netherlands)
it-IT — Italian (Italy)
pl-PL — Polish (Poland)
pt-BR — Portuguese (Brazil)
pt-PT — Portuguese (Portugal)
ja-JP — Japanese (Japan)
tr-TR — Turkish (Turkey)
vi-VN — Vietnamese (Vietnam)
id-ID — Indonesian (Indonesia)
th-TH — Thai (Thailand)
ro-RO — Romanian (Romania)
el-GR — Greek (Greece)
cs-CZ — Czech (Czech Republic)
fi-FI — Finnish (Finland)
hi-IN — Hindi (India)
bg-BG — Bulgarian (Bulgaria)
da-DK — Danish (Denmark)
he-IL — Hebrew (Israel)
ms-MY — Malay (Malaysia)
fa-IR — Persian (Iran)
sk-SK — Slovak (Slovakia)
sv-SE — Swedish (Sweden)
hr-HR — Croatian (Croatia)
tl-PH — Tagalog (Philippines)
hu-HU — Hungarian (Hungary)
no-NO — Norwegian (Norway)
sl-SI — Slovenian (Slovenia)
ca-ES — Catalan (Spain)
nn-NO — Norwegian Nynorsk (Norway)
ta-IN — Tamil (India)
af-ZA — Afrikaans (South Africa)

Note

When you select the TTS Max model and enter text in a language that is not supported, FineVoice will automatically use the FineVoice TTS model for conversion. To ensure accurate pronunciation and naturalness, this process may take more time.

Part 2. Quick Start

FineVoice Text to Speech lets you easily explore all TTS features, find the ideal voice for your project, test different texts and effects, and save or share your creations.

Step 1. Go to FineVoice

Select Text to Speech in the left menu bar.

Step 2. Enter Text

Input the text you want to convert to speech. For a quick start, you can use the default text provided at the bottom. You can also add emotion and style tags to your text.

Step 3. Select an AI Voice

Click "AI Voice" on the right to browse all FineVoice voices. Search by type, age, gender, or keywords, and preview each voice before selecting.

Step 4. Choose a Voice Generation Model

Select the TTS model that suits your needs. For best results, use the advanced TTS Max model.

Step 5. Generate Speech

Click the "Generate" button in the center. Once the audio is generated, it will play automatically. You can find all previously generated voices in the "History" section.

Note

For greater control over your audio, adjust advanced settings like Pitch, Speed, Temperature, and Top P. Use emotion and style tags to refine emotional expression, style, and non-verbal elements of the speech.

Part 3. Notes and Instructions

Emotion and Expression

Emotion and style tags control how your text is read aloud. For best results, place them at the beginning of your text to apply to the following content.

Emotion: [happy] [sad] [angry] [surprised] [fearful] [disgusted]
Delivery Style: [laughing] [whispering]
Vocalization: [breathe] [clear_throat] [cough] [laugh] [sigh] [yawn]

Example:

[happy] I can't believe this is happening.
[clear_throat] Did you hear what I said? [sigh] You never listen to me!

For best results, use only one emotion or style tag at the beginning of your text. Using multiple tags or placing them mid-text may cause unexpected results. To apply different emotions or styles, split your text into segments and add the tag at the start of each.

You can use multiple non-verbal tags within a single text to add sound effects throughout the speech.

Custom Pronunciation

Sometimes you need a word pronounced a specific way—especially uncommon words like company names, brand names, nicknames, geographic locations, medical or legal terms. Custom pronunciation lets you precisely control how these words are spoken.

FineVoice TTS Max supports inline IPA phoneme symbols for easy custom pronunciation. Please use the International Phonetic Alphabet (IPA) format, enclosed in slashes (/ /). For example:

Suppose you're building an AI travel agency that recommends the destination “Crete”, which is pronounced in English as /kriːt/ (“kreet”).
You can ensure the correct pronunciation by passing it inline: Your interests are a perfect match for a honeymoon in /kriːt/.

The model will replace any inline IPA pronunciations in your text. If your text is generated by an LLM, simply replace the original spelling with the IPA notation before sending it to the TTS model.

Find the Correct IPA Phonemes

If you're unsure about the correct phonemes, there are several ways to find them:

Ask an LLM like ChatGPT. For example, you can ask:
"What is the IPA phoneme for the word ‘Crete,’ pronounced like ‘kreet’?"

Once you have the correct phonemes, you can embed them directly in your TTS request: Your adventure in /kriːt/ begins today.

Advanced Settings

Pitch
Adjust the pitch of the voice. Negative values make the voice deeper and richer, while positive values make it brighter and more lively.
Speed
Control the speaking speed. 1.0 is the normal speed, 0.5 is half of the normal speed, and 1.5 is 1.5 times the normal speed.
Temperature (Creativity Level)
Control the output randomness. Higher values produce more varied and expressive results; lower values make output more predictable. For stability, maintain the temperature between 0.6 and 1.
Top P (Sampling Range)
Control the probability distribution range during voice generation. Lower values yield more conservative, stable results; higher values allow broader sampling for more creative and diverse output. Recommended range: 0.8 to 1.

Part 4. Best Practices

General Best Practices

Recommended:	Not Recommended:
Use appropriate punctuation for natural pauses Add emotion tags for expressive speech Split long texts into paragraphs Use consistent formatting	Use all caps (except for SHOUTING) Randomly mix multiple languages Include unnecessary special characters Forget punctuation

1. Choose the Right Voice:

Different voices are suited to different applications. Select a voice that matches the emotional range and style you need. For example, choose a stable, calm voice for a meditation app, or a more expressive and energetic voice for a motivational fitness coach.

2. Match Voice to Text Language:

Speech synthesis works best when the text matches the native language of the selected voice. While cross-language synthesis is possible, matching the voice's native language to the text ensures the highest quality, pronunciation, and naturalness.

3. Normalize Complex Text:

If the model mispronounces complex phrases (e.g., phone numbers or currency amounts), normalize them for clarity. This is especially important for non-English languages. Examples include:

Phone numbers: “(123)456-7891” -> “one two three, four five six, seven eight nine one”
Dates: 5/6/2025 -> “may sixth twenty twenty five” (helpful since date formats may vary)
Times: “12:55 PM” -> “twelve fifty-five PM”
Emails: test@example.com -> “test at example dot com”
Monetary values: $5,342.29 -> “five thousand three hundred and forty two dollars and twenty nine cents”
Symbols: 2+2=4 -> “two plus two equals four”

4. Adjust Temperature:

Temperature controls the randomness of audio output. Higher values produce more expressive results, ideal for demo clips or sound effects. Lower values make output more predictable, but too low may reduce quality. For real-time use, we recommend keeping the temperature between 0.6 and 1, with 0.8 offering the most stability.

Emotion Tags

1. Choose Context-Appropriate Tags:

Tags work best when they match the text content. If a tag conflicts with the text, the model may struggle to process it correctly.

(happy) I just got the job! This is the best day ever!

The tag (happy) matches the context of "getting a job", making the speech sound natural.

2. Avoid Tag Conflicts:

When using multiple tags for a single text, make sure they do not conflict with each other.

[sad] I miss you so much. [laugh] Life just feels so empty without you.

3. Split Text:

Emotion and expression tags work best at the beginning of the text, with one tag per request. Using multiple tags or placing them mid-text may lead to unpredictable results.

[surprised] What? You actually finished the project ahead of schedule? [happy] That's unbelievable! I didn't think it was possible.

It's better to format the text like this:

[surprised] What? You actually finished the project ahead of schedule?
[happy] That's unbelievable! I didn't think it was possible.

4. Repeat Nonverbal Sounds When Needed:

If a nonverbal sound is often missed, repeating the tag can improve accuracy. This works best for naturally repeating sounds, like [laugh] [laugh] or [cough] [cough].

(laugh)(laugh) Oh, that was hilarious! I can't stop laughing!

Part 5. Frequently Asked Questions

1. How do I add pauses in the text?

You can insert spaces, commas, periods, or custom pause tags to create pauses in speech synthesis. For example:

Short pause: Use (break) or a comma.
Long pause: Use (long-break) or a line break.

2. How can I generate speech for text over 5,000 characters?

The current version supports up to 5,000 characters per request. For longer texts, split your content into smaller sections, generate speech for each part, and use FineCut to merge the audio files. This ensures smooth speech and avoids system limitations.

3. Why do community models sometimes have accents?

Most community models are created through voice cloning. If the source audio doesn't match the target language, accents may occur (e.g., cloning a Chinese voice from English audio can result in an English accent).
Solutions:

Use source audio in the target language.
For existing models, retraining or fine-tuning can help reduce accents.

4. Other Suggestions

Community models are intended for personal entertainment or learning. For commercial use, we recommend official commercial models.
You can adjust speed, volume, and emotion parameters during speech generation to improve naturalness and listenability.

Part 6. Support & Feedback

If you encounter any issues, use the Feedback button in the top right corner of the software.

You can also email our support team: support@fineshare.com
For help from the community, join our Discord: https://discord.com/invite/HquSyU4Dy8
Connect with us on social media: https://twitter.com/FineShare_Soft/ or https://x.com/finevoice_ai

FineVoice Text to Speech Guide

Recommended Posts for You

Start Your AI Voice Journey with FineVoice for Free