flow.waitForPipeline
Introduction
flow.waitForPipeline is a Flow Command that synchronizes a Scenario with the playback duration of an injected audio file.
Rather than relying on hardcoded delays, Flowstride calculates exactly how long the injected audio will play and pauses execution accordingly.
This ensures voice-driven applications receive the complete audio stream before the Scenario continues.
The Problem It Solves
When testing audio-driven pipelines, such as AI voice assistants, speech-to-text systems, customer support bots, or WebRTC applications, automation engines execute commands much faster than audio can be played.
For example, if you inject a 10-second audio file and immediately stop the recording, the application may only receive a fraction of the audio. This results in incomplete transcripts, failed assertions, or unreliable AI responses. Using fixed delays is equally problematic.
If the audio changes from 10 seconds to 5 seconds, the test unnecessarily waits an additional 5 seconds. If the audio becomes longer, the recording may stop before playback finishes.
flow.waitForPipeline eliminates this problem by dynamically calculating the playback duration of the injected audio file.
How flow.waitForPipeline Works
Flowstride uses Dynamic Byte-Rate Calculation to determine the playback duration of the currently injected audio file.
Rather than relying on hardcoded wait times, the framework reads the audio file itself and mathematically calculates how long it will take to play.
Header Extraction
Flowstride opens the injected .wav file and reads the first 44 bytes, which represent the standard RIFF/WAVE header.
Mathematical Duration Resolution
Using information stored within the header, Flowstride extracts the audio byte rate and calculates the playback duration.
This allows the framework to determine exactly how long the audio stream will take to complete.
Smart Thread Suspension
Once the playback duration has been calculated, Flowstride pauses execution for the calculated duration, minus a small 500 ms buffer.
This gives the application sufficient time to receive the audio stream while allowing the Scenario to continue smoothly once playback is nearly complete.
Syntax
flow.waitForPipeline;Parameters
This command does not accept any parameters.
Example
Wait for voice playback to complete
Feature: Voice Search
Scenario: Verify speech transcription
Given "Initialize the virtual microphone"
flow.injectAudio "fixtures/customer-question.wav";
When "Open the voice search page"
flow.open "/voice-search";
When "Start voice recording"
flow.click button "Start Recording";
When "Wait for the injected audio to finish playing"
flow.waitForPipeline;
Then "The spoken request is transcribed"
flow.expect transcript contains "find nearby coffee shops";When to Use flow.waitForPipeline
Use flow.waitForPipeline immediately after the application begins listening for microphone input.
Typical examples include:
- Speech-to-text applications
- Voice authentication
- AI voice assistants
- Voice search
- Customer support bots
- WebRTC voice workflows
Important Notes
flow.injectAudio must be executed first
flow.waitForPipeline calculates the playback duration of the currently injected audio file.
If no audio has been injected, Flowstride cannot determine the playback duration and the Scenario will fail.
WAV duration is calculated automatically
Flowstride reads the WAV header and computes the playback duration mathematically.
This allows execution to remain synchronized with the injected audio without requiring manual timing adjustments.
Common Mistakes
Do not stop recording immediately
Avoid stopping microphone recording immediately after starting it.
When "Start recording"
flow.click button "Start Recording";
When "Stop recording"
flow.click button "Stop";Instead, wait for the injected audio to finish playing.
When "Start recording"
flow.click button "Start Recording";
When "Wait for playback"
flow.waitForPipeline;
Then "Verify the transcript"
flow.expect transcript contains "find nearby coffee shops";Do not use with uninjected audio
flow.waitForPipeline only works with audio previously prepared using flow.injectAudio.