Add SoundInstruction::SpeakNumberWith

[benjhar]

Adds SoundInstruction::SpeakNumberWith, which allows us to pass a NumberSpeakerConfig to determine where to find voice files.

[Copilot]

🟡 Changes recommended

The speak_number_with deserialization path contains a confirmed bug/mismatch in required-field handling (and should be aligned with NumberSpeakerConfig’s derived deserialization).

Once you've addressed the issues Copilot identified, you can request another Copilot review.

Pull request overview

Adds serialization/deserialization support needed to introduce SoundInstruction::SpeakNumberWith, enabling callers to specify a NumberSpeakerConfig (voice file location/format and language) as part of the instruction JSON format.

Changes:

• Add SoundInstruction::SpeakNumberWith(i64, NumberSpeakerConfig) with JSON (de)serialization and a round-trip test.
• Introduce NumberSpeakerConfig (serde-enabled) with defaults that depend on the active system language.
• Implement serde::Serialize/serde::Deserialize for LanguageTag and add basic serde tests.

File summaries

File	Description
src/language.rs	Adds serde (de)serialization for LanguageTag plus tests.
src/audio/sound_instruction.rs	Adds SpeakNumberWith variant and JSON (de)serialization + test.
src/audio.rs	Introduces NumberSpeakerConfig (serde-enabled) and a deserialize test.
Cargo.toml	Enables serde derive feature.
Cargo.lock	Adds serde_derive dependency via serde derive feature.

Review details

• Files reviewed: 4/5 changed files
• Comments generated: 2
• Review effort level: Lite

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[benhansen-io]

What about having SpeakNumber take an Option and make its json representation allow for an optional "config" field?

[benjhar]

I was avoiding needing to update uses of SpeakNumber in activities, which there are a few of. I could still do this though if you think it's preferable. We could change the json name from speak_number_with to speak_number and then check for if config is set, which would reduce the number of json instructions while keeping variants separate.

[benhansen-io]

I think most activities use the SoundBuilder which we should be able to adapt in a backwards compatible way.

[benhansen-io]

Oh and to be clear I am talking about the SpeakNumber SoundInstruction. Not the SpeakNumber struct used in native activities.