When using the multiple generations API, i'm finding that breaking text mid-word produces bad resul

[jolly imp]

AI-generated Answer for tomkail

When using the next_text and previous_text parameters, it's important to maintain context between chunks of text. However, breaking text mid-word is not recommended, as it can lead to unnatural sound blending. These parameters should be used between complete sentences to achieve more natural-sounding results. For example, you can use them to connect full sentences rather than splitting a single sentence or word. This approach generally yields better outcomes.

[proud parcel]

IS this true?

[jolly imp]

AI-generated Answer for tomkail

When using the ElevenLabs Text-to-Speech API, breaking text mid-word can lead to unwanted audio artifacts or inconsistencies, even if you use the next_text and previous_text parameters. This issue can occur because the API may not handle mid-word breaks as seamlessly as complete words or sentences. It's generally advisable to avoid breaking text mid-word to ensure more consistent and reliable audio output. If you continue to experience issues, consider adjusting the text segmentation to align with natural word boundaries.

[proud parcel]

IT's not written anywhere in the docs

[jolly imp]

AI-generated Answer for tomkail

Breaking text mid-word can lead to poor results, as the API may not handle such breaks effectively. It's generally advisable to avoid splitting text in the middle of a word to ensure better audio quality. Instead, try to break the text at natural pauses, like sentence or paragraph boundaries, to maintain coherence and quality in the generated speech.

[proud parcel]

Where are these best practices written?