>- Stream call audio in real-time, fork media to external destinations, and transcribe speech live. Use for real-time analytics and AI integrations. This skill provides Python SDK examples.
npx skills add https://github.com/team-telnyx/ai --skill telnyx-voice-streaming-python
<!-- Auto-generated from Telnyx OpenAPI specs. Do not edit. -->
pip install telnyx
import os
from telnyx import Telnyx
client = Telnyx(
api_key=os.environ.get("TELNYX_API_KEY"), # This is the default and can be omitted
)
All examples below assume client is already initialized as shown above.
All API calls can fail with network errors, rate limits (429), validation errors (422),
or authentication errors (401). Always handle errors in production code:
import telnyx
try:
result = client.messages.send(to="+13125550001", from_="+13125550002", text="Hello")
except telnyx.APIConnectionError:
print("Network error — check connectivity and retry")
except telnyx.RateLimitError:
# 429: rate limited — wait and retry with exponential backoff
import time
time.sleep(1) # Check Retry-After header for actual delay
except telnyx.APIStatusError as e:
print(f"API error {e.status_code}: {e.message}")
if e.status_code == 422:
print("Validation error — check required fields and formats")
Common error codes: 401 invalid API key, 403 insufficient permissions,
404 resource not found, 422 validation error (check field formats),
429 rate limited (retry with exponential backoff).
Call forking allows you to stream the media from a call to a specific target in realtime. This stream can be used to enable realtime audio analysis to support a
variety of use cases, including fraud detection, or the creation of AI-generated audio responses. Requests must specify either the target attribute or the rx and tx attributes.
POST /calls/{call_control_id}/actions/fork_start
Optional: client_state (string), command_id (string), rx (string), stream_type (enum: decrypted), tx (string)
response = client.calls.actions.start_forking(
call_control_id="550e8400-e29b-41d4-a716-446655440000",
)
print(response.data)
Returns: result (string)
Stop forking a call. Expected Webhooks:
call.fork.stoppedPOST /calls/{call_control_id}/actions/fork_stop
Optional: client_state (string), command_id (string), stream_type (enum: raw, decrypted)
response = client.calls.actions.stop_forking(
call_control_id="550e8400-e29b-41d4-a716-446655440000",
)
print(response.data)
Returns: result (string)
Start streaming the media from a call to a specific WebSocket address or Dialogflow connection in near-realtime. Audio will be delivered as base64-encoded RTP payload (raw audio), wrapped in JSON payloads. Please find more details about media streaming messages specification under the link.
POST /calls/{call_control_id}/actions/streaming_start
Optional: client_state (string), command_id (string), custom_parameters (array[object]), dialogflow_config (object), enable_dialogflow (boolean), stream_auth_token (string), stream_bidirectional_codec (enum: PCMU, PCMA, G722, OPUS, AMR-WB, L16), stream_bidirectional_mode (enum: mp3, rtp), stream_bidirectional_sampling_rate (enum: 8000, 16000, 22050, 24000, 48000), stream_bidirectional_target_legs (enum: both, self, opposite), stream_codec (enum: PCMU, PCMA, G722, OPUS, AMR-WB, L16, default), stream_track (enum: inbound_track, outbound_track, both_tracks), stream_url (string)
response = client.calls.actions.start_streaming(
call_control_id="550e8400-e29b-41d4-a716-446655440000",
stream_url="wss://example.com/audio-stream",
)
print(response.data)
Returns: result (string)
Stop streaming a call to a WebSocket. Expected Webhooks:
streaming.stoppedPOST /calls/{call_control_id}/actions/streaming_stop
Optional: client_state (string), command_id (string), stream_id (uuid)
response = client.calls.actions.stop_streaming(
call_control_id="550e8400-e29b-41d4-a716-446655440000",
)
print(response.data)
Returns: result (string)
Start real-time transcription. Transcription will stop on call hang-up, or can be initiated via the Transcription stop command. Expected Webhooks:
call.transcriptionPOST /calls/{call_control_id}/actions/transcription_start
Optional: client_state (string), command_id (string), transcription_engine (enum: Google, Telnyx, Deepgram, Azure, A, B), transcription_engine_config (object), transcription_tracks (string)
response = client.calls.actions.start_transcription(
call_control_id="550e8400-e29b-41d4-a716-446655440000",
)
print(response.data)
Returns: result (string)
Stop real-time transcription.
POST /calls/{call_control_id}/actions/transcription_stop
Optional: client_state (string), command_id (string)
response = client.calls.actions.stop_transcription(
call_control_id="550e8400-e29b-41d4-a716-446655440000",
)
print(response.data)
Returns: result (string)
Telnyx signs webhooks with Ed25519. Each request includes telnyx-signature-ed25519
and telnyx-timestamp headers. Always verify signatures in production:
# In your webhook handler (e.g., Flask — use raw body, not parsed JSON):
@app.route("/webhooks", methods=["POST"])
def handle_webhook():
payload = request.get_data(as_text=True) # raw body as string
headers = dict(request.headers)
try:
event = client.webhooks.unwrap(payload, headers=headers)
except Exception as e:
print(f"Webhook verification failed: {e}")
return "Invalid signature", 400
# Signature valid — event is the parsed webhook payload
print(f"Received event: {event.data.event_type}")
return "OK", 200
The following webhook events are sent to your configured webhook URL.
All webhooks include telnyx-timestamp and telnyx-signature-ed25519 headers for Ed25519 signature verification. Use client.webhooks.unwrap() to verify.
| Event | Description |
|-------|-------------|
| callForkStarted | Call Fork Started |
| callForkStopped | Call Fork Stopped |
| callStreamingFailed | Call Streaming Failed |
| callStreamingStarted | Call Streaming Started |
| callStreamingStopped | Call Streaming Stopped |
| transcription | Transcription |
callForkStarted
| Field | Type | Description |
|-------|------|-------------|
| data.record_type | enum: event | Identifies the type of the resource. |
| data.event_type | enum: call.fork.started | The type of event being delivered. |
| data.id | uuid | Identifies the type of resource. |
| data.occurred_at | date-time | ISO 8601 datetime of when the event occurred. |
| data.payload.connection_id | string | Call Control App ID (formerly Telnyx connection ID) used in the call. |
| data.payload.call_control_id | string | Unique ID for controlling the call. |
| data.payload.call_leg_id | string | ID that is unique to the call and can be used to correlate webhook events. |
| data.payload.call_session_id | string | ID that is unique to the call session and can be used to correlate webhook events. |
| data.payload.client_state | string | State received from a command. |
| data.payload.stream_type | enum: decrypted | Type of media streamed. |
callForkStopped
| Field | Type | Description |
|-------|------|-------------|
| data.record_type | enum: event | Identifies the type of the resource. |
| data.event_type | enum: call.fork.stopped | The type of event being delivered. |
| data.id | uuid | Identifies the type of resource. |
| data.occurred_at | date-time | ISO 8601 datetime of when the event occurred. |
| data.payload.connection_id | string | Call Control App ID (formerly Telnyx connection ID) used in the call. |
| data.payload.call_control_id | string | Unique ID for controlling the call. |
| data.payload.call_leg_id | string | ID that is unique to the call and can be used to correlate webhook events. |
| data.payload.call_session_id | string | ID that is unique to the call session and can be used to correlate webhook events. |
| data.payload.client_state | string | State received from a command. |
| data.payload.stream_type | enum: decrypted | Type of media streamed. |
callStreamingFailed
| Field | Type | Description |
|-------|------|-------------|
| data.record_type | enum: event | Identifies the resource. |
| data.event_type | enum: streaming.failed | The type of event being delivered. |
| data.id | uuid | Identifies the type of resource. |
| data.occurred_at | date-time | ISO 8601 datetime of when the event occurred. |
| data.payload.call_control_id | string | Call ID used to issue commands via Call Control API. |
| data.payload.connection_id | string | Call Control App ID (formerly Telnyx connection ID) used in the call. |
| data.payload.call_leg_id | string | ID that is unique to the call and can be used to correlate webhook events. |
| data.payload.call_session_id | string | ID that is unique to the call session and can be used to correlate webhook events. |
| data.payload.client_state | string | State received from a command. |
| data.payload.failure_reason | string | A short description explaning why the media streaming failed. |
| data.payload.stream_id | uuid | Identifies the streaming. |
| data.payload.stream_type | enum: websocket, dialogflow | The type of stream connection the stream is performing. |
callStreamingStarted
| Field | Type | Description |
|-------|------|-------------|
| data.record_type | enum: event | Identifies the type of the resource. |
| data.event_type | enum: streaming.started | The type of event being delivered. |
| data.id | uuid | Identifies the type of resource. |
| data.occurred_at | date-time | ISO 8601 datetime of when the event occurred. |
| data.payload.call_control_id | string | Call ID used to issue commands via Call Control API. |
| data.payload.connection_id | string | Call Control App ID (formerly Telnyx connection ID) used in the call. |
| data.payload.call_leg_id | string | ID that is unique to the call and can be used to correlate webhook events. |
| data.payload.call_session_id | string | ID that is unique to the call session and can be used to correlate webhook events. |
| data.payload.client_state | string | State received from a command. |
| data.payload.stream_url | string | Destination WebSocket address where the stream is going to be delivered. |
callStreamingStopped
| Field | Type | Description |
|-------|------|-------------|
| data.record_type | enum: event | Identifies the type of the resource. |
| data.event_type | enum: streaming.stopped | The type of event being delivered. |
| data.id | uuid | Identifies the type of resource. |
| data.occurred_at | date-time | ISO 8601 datetime of when the event occurred. |
| data.payload.call_control_id | string | Call ID used to issue commands via Call Control API. |
| data.payload.connection_id | string | Call Control App ID (formerly Telnyx connection ID) used in the call. |
| data.payload.call_leg_id | string | ID that is unique to the call and can be used to correlate webhook events. |
| data.payload.call_session_id | string | ID that is unique to the call session and can be used to correlate webhook events. |
| data.payload.client_state | string | State received from a command. |
| data.payload.stream_url | string | Destination WebSocket address where the stream is going to be delivered. |
transcription
| Field | Type | Description |
|-------|------|-------------|
| data.record_type | enum: event | Identifies the type of the resource. |
| data.event_type | enum: call.transcription | The type of event being delivered. |
| data.id | uuid | Identifies the type of resource. |
| data.occurred_at | date-time | ISO 8601 datetime of when the event occurred. |
| data.payload.call_control_id | string | Unique identifier and token for controlling the call. |
| data.payload.call_leg_id | string | ID that is unique to the call and can be used to correlate webhook events. |
| data.payload.call_session_id | string | ID that is unique to the call session and can be used to correlate webhook events. |
| data.payload.client_state | string | Use this field to add state to every subsequent webhook. |
| data.payload.connection_id | string | Call Control App ID (formerly Telnyx connection ID) used in the call. |
Python library for working with DICOM (Digital Imaging and Communications in Medicine) files. Use this skill when reading, writing, or modifying medical imaging data in DICOM format, extracting pixel data from medical images (CT, MRI, X-ray, ultrasound), anonymizing DICOM files, working with DICOM metadata and tags, converting DICOM images to other formats, handling compressed DICOM data, or processing medical imaging datasets. Applies to tasks involving medical image analysis, PACS systems, radiology workflows, and healthcare imaging applications.
Python library for working with DICOM (Digital Imaging and Communications in Medicine) files. Use this skill when reading, writing, or modifying medical imaging data in DICOM format, extracting pixel data from medical images (CT, MRI, X-ray, ultrasound), anonymizing DICOM files, working with DICOM metadata and tags, converting DICOM images to other formats, handling compressed DICOM data, or processing medical imaging datasets. Applies to tasks involving medical image analysis, PACS systems, radiology workflows, and healthcare imaging applications.
Best practices for Remotion - Video creation in React
Generate AI-powered podcast-style audio narratives using Azure OpenAI's GPT Realtime Mini model via WebSocket. Use when building text-to-speech features, audio narrative generation, podcast creation from content, or integrating with Azure OpenAI Realtime API for real audio output. Covers full-stack implementation from React frontend to Python FastAPI backend with WebSocket streaming.
Best practices for Remotion - Video creation in React
Port an existing Remotion (React) composition''s source to HyperFrames HTML. Use ONLY on an explicit ask to port/convert/migrate/translate a Remotion source — one-way, Remotion-only. A passing Remotion mention, reference-only code, or "make something like my Remotion video" is a fresh build (/general-video). Unclear → /hyperframes.
Use this skill when building applications with Gemini API hosted models, including Gemini and Gemma 4, working with multimodal content (text, images, audio, video), implementing function calling, using structured outputs, or needing current model specifications. Covers SDK usage...
Turn error logs, screenshots, voice notes, and rough bug reports into crisp, developer-ready GitHub issues with repro steps, impact, and evidence.
Take team-telnyx/telnyx-voice-streaming-python from the repository into ~/.claude/skills for personal
use, or into .claude/skills inside a project.
The agent identifies a skill by the name field in its header. Two skills with the
same name cannot sit side by side — one of them will be ignored.
The instructions reference pip.
Without those the skill loads but fails at the first command.