Voices

About AI voices

AI voices are created by training machine learning algorithms on recorded voices of professional voice actors. Hence they can transform the text into human-sounding voice-overs using speech synthesis.

Murf Studio presently supports voices in 20 languages, segmented into accents and age categories, for a total of 120+ different voice selections.

Languages

Voices and Plans

Murf's text to speech voices are split into two categories: Basic Voices and Pro Voices.

Basic Voices

These voices can be accessed with the Basic plan (monthly/annual), Pro Plan (monthly/annual) & Enterprise Plan (monthly/annual).
The basic voices consist of only 60 voice actors across ten languages.

The ten languages are as follows:

English
French
German
Italian
Spanish
Portuguese
Russian
Hindi
Arabic
Tamil

Pro Voices

These voices can be accessed with the Pro plan (monthly/annual) and the Enterprise plan (monthly/annual) Only
Pro and Enterprise users can access both Basic and Pro voices.
The Pro voices consist of all 120+ voices across 20 languages available in the Studio.

Accents

At Murf, we understand that every language may have different dialects, and it is essential to convey them perfectly.

The following languages support multiple text to speech accents: English, French, Spanish, Portuguese and Chinese

English	US & Canada UK Australia India Kids
French	France Canada
Spanish	Mexico Spain
Portuguese	Brazil Portugal
Chinese	Simplified Cantonese Taiwanese

Age Group

Murf Studio currently has voices across the following age groups:

Young Adult: A young adult's voice can be attributed to up-beat, energetic, or enthusiastic voice-overs. Eighty-eight of our AI voices support this version across 20 Languages. Out of these, 39 are male and 49 female voices.

Middle-Aged: Voices have slightly deeper tones. We have 54 middle-aged voices available across 20 languages. Out of these, 28 are male, and 26 are female voices.

Children: We have three children's voices available in English (Kids), out of which two are female and one male voice.

Voice Tone

Users can use this feature to produce a more realistic voice over based on the video's theme or target demographic.

The following voices support tones:

English (UK)
- Eliza- General and Newscast (Basic Voice)
English (US & Canada)
- Ava- General, Newscast formal, Newscast casual, Narration-professional, Chat, Customer service, Cheerful and Empathetic (Basic Voice)
- Adam- General Newscast and conversational (Basic Voice)
- Nate- General and Newscast (Basic Voice)
- Cathy- General, Assistant, Chat, Customer service, and newscast (Pro Voice)
- Jessica- General Newscast and conversational (Pro Voice)
- Rachel- General, cheerful, angry, and sad (Pro Voice)
- Ken- Conversational and Promo (Pro Voice)
- Samantha- Coversational, and Luxury (Pro Voice)
- Julie- Coversational, and Cheerful (Pro Voice)
Chinese (Simplified)
- Liu- General, Assistant, chat, customer service, Newscast, affectionate, Angry, Calm, Cheerful, Disgruntled, fearful, gentle, lyrical, sad, serious (Pro Voice)
- Chao- General, Embarrassed, Calm, Fearful, Cheerful, Disgruntled, Serious, Sad, and angry (Pro Voice)
- Hong- Customer service, Narration-professional, Newscast- casual (Pro Voice)
Japanese
- Sakura- Chat, Customer service, and Cheerful (Pro Voice)
Portuguese (Brazil)
- Marcia- General and Calm

Pitch

In speech, the ear perceives a tone's relative highness or lowness. This is based on the pitch of the speech. You can set the desired pitch for your voice-over using this option. (High pitch or low pitch)

To modify the pitch to suit your voice-over needs, follow the instructions below.

Click on the text block to which the setting is to be applied.
Head to the 'Voices' tab and click on 'Pitch'.
Lower the pitch below 1.0x for a deeper effect.
Similarly, increase the pitch above 1.0x for a sharp or cartoonish effect.
Once updated, click on the generate button to generate audio and listen to the updated AI voice.

Speed

This option lets you control the speed at which the AI voice reads the script.

Click on the text block to which the setting is to be applied.
Head to the 'Voices' tab and click on 'Speed'.
For a slower pace, move the scale below 1.0x. Similarly, for a faster pace, move the scale above 1.0x
Once updated, click on the generate button to generate audio and listen to the updated AI voice.

Volume

Volume levels of the script can be increased or decreased using this option. The volume setting can stress a sentence with a greater volume or tell the sentence softly.

Loudness

The loudness tool helps control the voice-over volume and the background music individually.

Note:

LUFS stands for Loudness Unit to Full Scale (i.e., the maximum level a system can handle)
Range for each scale is 0% - 100%.
0% will mute the audio, and 100% will generate the highest loudness.

Auto Ducking

Auto Duck reduces (ducks) the volume of the background music whenever the system recognizes the voice-over being played. This feature offers a clear and crisp voice-over output, even when a background track is being played.

Emphasis

Adding intensity of expression for a given word or words can be done using the Emphasis tool. Adding emphasis increases the intonation of the voice at the point where the node has been placed.

Nodes are points in the emphasis graph which help you move the line of Emphasis up or down.
Each click in the graph will create a new 'Node' which you can use to manipulate the graph. (A maximum of 5 nodes can be created within a Graph for each selection)
The X-axis in the Graph represents the words available in the paragraph/text block.
The Y-axis represents the Emphasis you would like to apply to range between -100% being the lowest and 100% being the highest. The higher the graph wave set on the word corresponding in X-Axis, the higher the word's Emphasis.
The graph line represented in the middle should be used to plot the Emphasis range.

Pause

This is one of the essential tools used in syncing the voice-over and the video or PowerPoint presentation. Adding pauses also aids in increasing the overall realism of the voice over.

These are the following methods to add a pause in a project:

- Pre-defined pauses: There will be an Add Pause button from the voice tab in the top menu. Here you will find the following pre-defined pauses.
- Extra Weak (250ms), Weak (0.5s), Medium (0.75s), Strong (1s), Extra Strong (1.2s),
- Or enter a custom Pause time up to 5 seconds at once.
- There is also a 'No Pause' option where you can remove any pause in the voice-over. It creates a rushing effect between the words and where the pause is placed.
- Punctuation marks: Certain punctuation marks such as full stops (.) and commas(,) will add a natural pause to the sentence to make the output more accurate.
- Increasing voice block size: The options mentioned above add pauses between and at the end of sentences.
- You can also add longer pauses at the end of each voice block by selecting it from the bottom timeline and increasing the size.

The duration of the voice over and the block size can be found in the timer mentioned.

The time of the text block has increased after dragging the text block in the timeline.

How to use 'Pause' to add Introductory Music? How to start voice-over from the 'x' timestamp?

The first block of the project is to be left blank and start your script from the second block.
Then select the first block from the timeline, stretch it, and increase the size to 'x.'
Use the Build button for changes to reflect and preview your project.

Pronunciation

The Pronunciation tool allows you to alter the pronunciation of any word present in the project. This can be utilized when a specific voice cannot accuratelyt enunciate a word.

This feature can be found under the Voices tab in the top menu.

There are two different methods to update a word pronunciation:

1. Alternative Spelling
  - Breaking down the word into syllables and entering an alternative spelling of the word can help fix the pronunciation.
  - Eg
    - Murf > Murph
    - Present > Pre-zent
2. IPA
  - The International Phonetic Alphabet (IPA) is an alphabetic system of phonetic notation based primarily on the Latin script. It was devised by the International Phonetic Association in the late 19th century as a standardized representation of speech sounds in written form.
  - In phonology and linguistics, a phoneme is a unit of sound that can distinguish one word from another in a particular language.
  - IPA Phonemes can be used in Murf Studio to help arrive at accurate pronunciations of words.
  - Eg:
    - Recoil: riˈkɔɪl
    - Specific: spəsˈɪfɪk

Reference links to find IPA:

Select words to modify their pronunciation:

The selected sentence will be displayed. Click on the word to be updated.

Import: Shared Pronunciaiton Library

Common pronunciations can be saved and shared with team members. (Pro and Enterprise feature)
Once it is saved to the team library, any team member linked to the Admin account can access the pronunciation library and use the previously saved pronunciation.
To save an IPA, click on the icon present next to the word.

Advance toggle:

ON: When adding IPA pronunciation, ensure the toggle is ON for the Phonemes option to be available
OFF: When adding only alternative spelling, this toggle can be off.

Word:

The word is chosen from the sentence to update the pronunciation.

Pronunciation:

Enter either an IPA or alternative spelling.

Phonemes:

ON: For the system to read the entered pronunciation as an IPA, this needs to be toggled ON.
OFF: When entering alternative spelling

Project Level toggle:

ON: The pronunciation added would be applied for that word present anywhere in the project.
OFF: The pronunciation would only be applied at a sentence level.

Important Notes:

After entering the pronunciation, it can be previewed using the generate button.
- For IPA to be applied in the project, you must click on Save. Failed to generate: This would occur if the system does not recognise the IPA entered.
- Try an alternative IPA
- Reach out to the support team via email at support@murf.ai or on live chat to create an IPA
- If an unknown IPA is saved, your project will fail to synthesize
- After saving IPA, the sentences containing the word will need to be re-rendered.