#can the TTS script include mixture of english and devnagri script . Currently, i notice that non-

1 messages · Page 1 of 1 (latest)

tired radishBOT
#
AI-generated Answer for therealpankajupreti

The system does not officially support mixing languages for professional voice cloning. However, some users have had success using a combination of Hindi and English. You can try this approach, but if you encounter a language mismatch error, the support team can delete the voice clone to allow you to try again with a single language.

spare flint
#

i am not referring to voice cloneing stage, i am referring to usage while creating text to speech and using the already cloned voice

tired radishBOT
#
AI-generated Answer for therealpankajupreti

Currently, the system does not officially support mixing languages in text-to-speech scripts, which can lead to pronunciation issues with non-English words like Hindi. This is a limitation of the current setup.

To improve pronunciation, you might consider creating a voice clone from a native speaker of the target language or experimenting with different settings to enhance accuracy. However, mixing scripts directly in TTS might not yield the desired results due to the system's limitations.

spare flint
#

the voice clone is already of a native speaker, who has provided speech recordings in english. However as i understand the model is multi lingual.. so that same voice should be able to produce good accent in other languages

tired radishBOT
#
AI-generated Answer for therealpankajupreti

When using a cloned voice for text-to-speech, the AI model determines the language based on the input text. If the text includes a mix of English and Devanagari script, the AI might struggle with pronunciation, especially if the voice was cloned using English recordings.

For better results, it's recommended to use a voice that was cloned with samples in the target language or accent. If the cloned voice is primarily in English, it may not handle non-English words or scripts as effectively.

To improve pronunciation in other languages, consider using a voice that was specifically cloned with samples in those languages. You can also explore the Voice Library for voices that match the desired language and accent.

spare flint
#

is it possible to record training samples in english and say things like... t"he word moon in hindi pronounced as chandra" and while recording the traing the pronunciation of the supplied hindi word is in perfect accent

tired radishBOT
#
AI-generated Answer for therealpankajupreti

When using text-to-speech with a cloned voice, the model's ability to handle mixed languages, such as English and Hindi, can be limited. While the model is multilingual, it may not always produce perfect pronunciation for non-English words if the training samples were primarily in English.

To improve pronunciation, you might consider:

  1. Phonetic Spelling: Try using phonetic spelling for Hindi words in the text input to guide the pronunciation more accurately.

  2. Separate Language Samples: If possible, provide separate samples in Hindi to enhance the model's ability to handle Hindi words.

  3. Experiment with Text Input: Adjust the text input to see if different spellings or structures yield better pronunciation results.

These steps might help improve the pronunciation of Hindi words when using a voice primarily trained in English.

spare flint
#

i am now referring back to the training voice and need help to prepare the training voice samples. Knowkning that the supplied samples are all english, is it allowed to also include words that are hindi , just so the training model can correctly pronouce hindi words when using TTS