Documenting this because I could not find anyone else reporting it and it took me a week to track down.
On Genesys Cloud Voice (not BYOC), when an agent performs a blind transfer to another queue, the call recording has a 2-second audio gap at the exact moment of transfer. Both the customer audio and agent audio drop to silence for approximately 1.8-2.2 seconds, then resume when the new agent connects.
The live call has no audio gap - the customer hears hold music during the transfer and the transition is seamless. It is only the recording that has the gap.
I verified this across 50 transferred calls. The gap is present on 100% of blind transfers. Attended transfers do NOT have the gap. Consultative transfers do NOT have the gap. Only blind transfers.
The gap correlates with the recording file structure. When I download the recording via the API and inspect it, blind transfers create TWO separate audio segments that are stitched together with a gap. Attended transfers create a single continuous segment.
This matters for our QM team because the gap often cuts off the last word the first agent says before transferring, making it impossible to evaluate whether the agent properly announced the transfer to the customer.
Anyone else seeing this on GC Voice?
Yes. This is a known artifact of how GC Voice handles recording during blind transfers and it drives me absolutely insane for Salesforce activity logging.
The root cause: during a blind transfer, the GC Voice media server tears down the first call leg and establishes a new one. The recording service treats these as two separate media streams and creates two recording segments. When the segments are stitched for playback, there is a gap because the first segment ends when the BYE is sent and the second segment starts when the new INVITE is answered.
There is no configuration to fix this on the Genesys Cloud Voice side. The recording architecture fundamentally treats blind transfers as two separate call legs.
Workaround for QM: switch your agents to “Consult Transfer” instead of “Blind Transfer.” Consult transfer maintains a single call leg throughout the transfer process, which produces a continuous recording without gaps. The agent does not actually need to speak to the receiving queue - they can immediately complete the consult transfer without waiting. The UX is almost identical to blind transfer but the recording stays intact.
You can enforce this by removing the “Blind Transfer” permission from the agent role and only granting “Consult Transfer.”
The consult transfer workaround is what we use too. Quick Python snippet if you want to programmatically identify all recordings with blind transfer gaps for retroactive QM review:
import PureCloudPlatformClientV2 as purecloud
recording_api = purecloud.RecordingApi()
conversation_api = purecloud.ConversationsApi()
# Query conversations with blind transfers
body = purecloud.ConversationQuery()
body.interval = '2025-05-01T00:00:00Z/2025-05-15T00:00:00Z'
body.segment_filters = [{
'predicates': [{
'dimension': 'transferType',
'value': 'blind'
}]
}]
results = conversation_api.post_analytics_conversations_details_query(body)
for conv in results.conversations:
recordings = recording_api.get_conversation_recordings(conv.conversation_id)
if len(recordings) > 1:
print(f'Gap detected: {conv.conversation_id} - {len(recordings)} segments')
This flags conversations where the recording was split into multiple segments due to blind transfer.
Confirming same behavior on our GC Voice + Bold360 integrated environment. The consult transfer approach eliminates the gap.
One more detail: if you use the recording export API to download recordings for archival, blind transfer recordings with multiple segments are exported as separate audio files. You need to concatenate them yourself if your archival system expects a single file per conversation. The segments are ordered by the startTime field in the recording metadata.