Compare results
Running both models lets us compare our output with the corresponding provider’s output.
Lower costs · transparent model selection
Accessible AI.
An open explanation.
OpenSpeaker runs the corresponding model from your selected provider alongside our own model. We compare and score the results, then return the higher-scoring output. These comparisons also guide improvements to our models, helping us offer AI at a lower cost.
From your request to your result
Two models in parallel. One selected output.
Both branches receive the input. Their results meet at the comparison and scoring stages before an output is selected.
- 01Your inputYour request and selected model
- 02 / Run in parallel
- Provider modelCorresponding to your selection
- OpenSpeaker modelOur in-house model
- 03CompareBring both results together
- 04ScoreEvaluate the results
- 05Final outputReturn the higher-scoring result
Comparison and scoring → improvements to OpenSpeaker’s models
The process helps us identify where our models need to improve and make them more efficient over time.
Why this supports lower costs
Better models. More efficient AI.
Lower pricing is part of a continuing process of evaluating and improving our own models.
Improve our models
Comparison and scoring help us identify gaps and decide what to improve next.
Optimize costs over time
Improving the quality and efficiency of our models helps us maintain accessible pricing.
A quality goal, assessed by the system.
We aim for output that is comparable to or better than the selected provider’s result. A higher score guides selection within our system; it does not guarantee a better result for every use case or preference.
Common questions
Before you choose a model.
Will the output always come from the provider I choose?
No. Your selection determines the corresponding provider model in the parallel process. The final output is selected by score, so it may come from OpenSpeaker’s own model if that result scores higher.
Does this apply only to text to speech?
No. This parallel comparison, scoring and output-selection process applies to all AI models listed on the OpenSpeaker site.
How does this help keep prices low?
The comparisons guide improvements to OpenSpeaker’s models. Improving our own models and their operating efficiency helps us optimize costs over time and offer more accessible pricing.
Does a higher score guarantee better quality?
A higher score is the system’s basis for selecting the output. The goal is comparable or better quality, but results can vary with the request and your expectations. It does not guarantee identical output to a direct request to the selected provider.
Know the process. Choose with confidence.
Explore the tools and pricing with a clear understanding of how the final output is selected.