docs: update frame type names from VAD* to User*

Updates documentation to reflect the frame type rename in pipecat PR #4198:
- VADUserStartedSpeakingFrame → UserStartedSpeakingFrame
- VADUserStoppedSpeakingFrame → UserStoppedSpeakingFrame

Updated pages:
- guides/learn/speech-input.mdx
- server/utilities/observers/user-bot-latency-observer.mdx
- server/services/stt/soniox.mdx
- server/services/stt/sarvam.mdx
- server/frames/system-frames.mdx

Author: github-actions[bot]

api-reference/server/frames/system-frames.mdx

@@ -153,9 +153,9 @@ Broadcast when all active [user mute strategies](/api-reference/server/utilities
 
 ## VAD Events
 
-These frames are emitted directly by the Voice Activity Detection (VAD) processor and carry timing metadata. Higher-level speaking-state frames (`UserStartedSpeakingFrame`, `UserStoppedSpeakingFrame`) are derived from these.
+These frames are emitted directly by the Voice Activity Detection (VAD) processor and carry timing metadata. Higher-level speaking-state frames are derived from these.
 
-### VADUserStartedSpeakingFrame
+### UserStartedSpeakingFrame
 
 VAD confirmed that speech has started.
 
@@ -167,7 +167,7 @@ VAD confirmed that speech has started.
   Wall-clock time when the frame was created.
 </ParamField>
 
-### VADUserStoppedSpeakingFrame
+### UserStoppedSpeakingFrame
 
 VAD confirmed that speech has ended.
 

api-reference/server/services/stt/sarvam.mdx

@@ -180,7 +180,7 @@ stt = SarvamSTTService(
 
 - **Supported languages**: Bengali (bn-IN), Gujarati (gu-IN), Hindi (hi-IN), Kannada (kn-IN), Malayalam (ml-IN), Marathi (mr-IN), Tamil (ta-IN), Telugu (te-IN), Punjabi (pa-IN), Odia (od-IN), English (en-IN), and Assamese (as-IN).
 - **Model-specific parameter validation**: The service validates that parameters are compatible with the selected model. For example, `prompt` is not supported with `saarika:v2.5`, and `language` is not supported with `saaras:v2.5` (which auto-detects language).
-- **VAD modes**: When `vad_signals=False` (default), the service relies on Pipecat's local VAD and flushes the server buffer on `VADUserStoppedSpeakingFrame`. When `vad_signals=True`, the service uses Sarvam's server-side VAD and broadcasts speaking frames from the server.
+- **VAD modes**: When `vad_signals=False` (default), the service relies on Pipecat's local VAD and flushes the server buffer on `UserStoppedSpeakingFrame`. When `vad_signals=True`, the service uses Sarvam's server-side VAD and broadcasts speaking frames from the server.
 
 <Tip>
   The `InputParams` / `params=` pattern is deprecated as of v0.0.105. Use

api-reference/server/services/stt/soniox.mdx

@@ -113,7 +113,7 @@ Before using Soniox STT services, you need:
 </ParamField>
 
 <ParamField path="vad_force_turn_endpoint" type="bool" default="True">
-  Listen to `VADUserStoppedSpeakingFrame` to send a finalize message to Soniox.
+  Listen to `UserStoppedSpeakingFrame` to send a finalize message to Soniox.
   When enabled, Pipecat's local VAD triggers transcript finalization. When
   disabled, Soniox detects the end of speech natively.
 </ParamField>

api-reference/server/utilities/observers/user-bot-latency-observer.mdx

@@ -72,8 +72,8 @@ When tracing is enabled, latency measurements are automatically recorded as `tur
 The observer tracks conversation flow through these key events:
 
 1. **Client connects** (`ClientConnectedFrame`) → Records timestamp for first-bot-speech measurement
-2. **User starts speaking** (`VADUserStartedSpeakingFrame`) → Resets latency tracking
-3. **User stops speaking** (`VADUserStoppedSpeakingFrame`) → Records timestamp, accounting for VAD `stop_secs` delay
+2. **User starts speaking** (`UserStartedSpeakingFrame`) → Resets latency tracking
+3. **User stops speaking** (`UserStoppedSpeakingFrame`) → Records timestamp, accounting for VAD `stop_secs` delay
 4. **Bot starts speaking** (`BotStartedSpeakingFrame`) → Calculates latency and emits `on_latency_measured` and `on_latency_breakdown` events
 
 When `enable_metrics=True` in `PipelineParams`, the observer also collects per-service metrics (TTFB, text aggregation, function call latency) from `MetricsFrame` instances and includes them in the latency breakdown.

pipecat/learn/speech-input.mdx

@@ -98,8 +98,8 @@ While VAD detects speech vs. silence, it can't understand linguistic context. A
 2. **Turn End**: When the stop strategy determines the user is done, it emits `UserStoppedSpeakingFrame`
 
 <Note>
-  VAD also emits its own frames (`VADUserStartedSpeakingFrame`,
-  `VADUserStoppedSpeakingFrame`) which indicate raw speech/silence detection.
+  VAD also emits its own frames (`UserStartedSpeakingFrame`,
+  `UserStoppedSpeakingFrame`) which indicate raw speech/silence detection.
   These are inputs to the turn strategies, not the final turn decisions.
 </Note>