Document enable_text_normalization SWML AI param#362
Open
Devon-White wants to merge 3 commits into
Open
Conversation
Add the enable_text_normalization param to the AI verb's params object and the AI params reference page. - TypeSpec: new AIParams field as an inline string-literal union (heard | spoken | both | true | on | false | off | none | SWMLVar), default "both"; regenerate SWMLObject.json. - Docs: ParamField in the Speech Recognition section plus an anchored values sub-section covering the modes, disable keywords, and per-language graceful-skip behavior. Defaults and accepted values follow the FreeSWITCH mod_openai source (default is "both", enabled). The supported-language matrix is not enumerated in source (runtime FAR grammars), so it is described generally rather than tabulated.
Lead with the benefit and plain before/after examples; drop internal jargon (NeMo, ITN/TN acronyms up front, calling.error event) from the ParamField, values table, and TypeSpec @doc. Regenerate SWMLObject.json.
Contributor
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
Documents the
enable_text_normalizationSWML AI verb param — NeMo-based text normalization between written and spoken forms during AI voice conversations.heard— Inverse Text Normalization (ITN) on ASR output before the LLM (twenty three dollars→$23)spoken— Text Normalization (TN) on LLM output before TTS ($23→twenty three dollars)both— both directions (default)Changes
specs/swml/calling/Methods/ai/ai_params.tsp): newAIParamsfield as an inline string-literal union (heard | spoken | both | true | on | false | off | none | SWMLVar), defaultboth.SWMLObject.json): regenerated viatsp compile(additive only).ai_params/index.mdx):ParamFieldin the Speech Recognition section + anchored values sub-section (modes, disable keywords, per-language graceful-skip behavior).Source fidelity
Default and accepted values follow the FreeSWITCH
mod_openaisource (session.c): default isboth(enabled), andtrue/on/false/off/nonekeywords are accepted alongside the three canonical modes.The originally-supplied 19-language support matrix is not included: supported languages are determined at runtime by which NeMo FAR grammar files are deployed (not enumerated in source), so the docs describe the lazy-load + graceful-skip behavior generally instead of publishing an unverified table.
auto_correct(mutually exclusive, but Partial/stub with an unwired OART TODO) is intentionally left undocumented.