TTS Pronunciation Accuracy via Transliteration and Phoneme Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing text-to-speech systems poorly pronounce foreign names using target language pronunciation rules, leading to a non-standard and inconsistent pronunciation of names in the target language alphabet, which differs significantly from the actual pronunciation.
Innovation Solution
The method identifies the language of origin for a named entity, transliterates it to the native script, and uses a grapheme-to-phoneme converter to generate a phoneme sequence, mapping it back to the target language phonemes to achieve accurate pronunciation, leveraging advancements in data mining and transliteration technologies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the TTS system uses target language pronunciation rules for foreign names, then the system maintains consistency in pronunciation rules, but the pronunciation accuracy of foreign names deteriorates
Solution Approach 1:
The patent introduces an intermediary process between the foreign name input and TTS output. This intermediary identifies the language of origin, performs transliteration to the target script, and uses phoneme mapping to preserve the original pronunciation characteristics while adapting to the target language TTS system, thereby resolving the conflict between pronunciation accuracy and language adaptability
Solution Approach 2:
The patent applies preliminary actions by pre-identifying the language of origin and pre-transliterating foreign names before they reach the TTS system. This preparation work includes creating phoneme sequences and mapping them to target language phonemes in advance, ensuring accurate pronunciation without requiring the TTS system to adapt to multiple language rules
2Measurement precision
If the TTS system uses transliteration to preserve original pronunciation, then pronunciation accuracy improves, but system complexity increases
Solution Approach 1:
The patent segments the name processing task into distinct components: language identification, transliteration to target script, phoneme sequence generation, and phoneme mapping. Each component is handled by a specialized module, making the overall complex system manageable and maintainable while achieving high pronunciation accuracy
Solution Approach 2:
The patent creates a universal phoneme mapping layer that works across different language pairs. The G2P converter and phoneme mapping mechanism serve multiple functions: they handle various source languages, work with different target languages, and can be applied to both proper nouns and common nouns, reducing overall system complexity through reusability
Data Source
AI summary
A system and method configured for use in a text-to-speech (TTS) system is provided. Embodiments may include identifying, using one or more processors, a word or phrase as a named entity and identifying a language of origin associated with the named entity. Embodiments may further include transliterating the named entity to a script associated with the language of origin. If the TTS system is operating in the language of origin, embodiments may include passing the transliterated script to the TTS system. If the TTS system is not operating in the language of origin, embodiments may include generating a phoneme sequence in the language of origin using a grapheme to phoneme (G2P) converter.