diff --git a/ovos_hivemind_solver/__init__.py b/ovos_hivemind_solver/__init__.py
index 13f7706..1627f3c 100644
--- a/ovos_hivemind_solver/__init__.py
+++ b/ovos_hivemind_solver/__init__.py
@@ -1,8 +1,8 @@
-from hivemind_bus_client import HiveMessageBusClient, HiveMessage, \
-    HiveMessageType
+from hivemind_bus_client import HiveMessageBusClient
 from ovos_bus_client.message import Message
 from ovos_plugin_manager.templates.solvers import QuestionSolver
 from ovos_utils.log import LOG
+from threading import Event
 
 
 class HiveMindSolver(QuestionSolver):
@@ -13,6 +13,8 @@ def __init__(self, config=None):
         config = config or {}
         super().__init__(config)
         self.hm = None
+        self._response = Event()
+        self._responses = []
         if self.config.get("autoconnect"):
             self.connect()
 
@@ -25,15 +27,46 @@ def connect(self):
         """
         self.hm = HiveMessageBusClient(useragent="ovos-hivemind-solver")
         self.hm.run_in_thread()
+        self.hm.on_mycroft("speak", self._receive_answer)
+        # NOTE: this message not yet introduced in ovos-core
+        self.hm.on_mycroft("ovos.utterance.handled", self._end_of_response)
+
+        ##############
+        # HACK - waiting for https://github.com/OpenVoiceOS/ovos-core/pull/478
+        # TODO - new bus message in ovos-core
+        #  to unambiguosly identify end of utterance parsing
+        #  currently these are 4 possible outcomes for any request
+        #  does not account for OCP pipeline requests
+        self.hm.on_mycroft("ovos.utterance.cancelled",
+                           self._end_of_response)
+        self.hm.on_mycroft("complete_intent_failure",
+                           self._end_of_response)
+        self.hm.on_mycroft("mycroft.skill.handler.complete",
+                           self._end_of_response)
+        self.hm.on_mycroft( "skill.converse.response",
+                           self._end_of_response)
+
+    def _end_of_response(self, message):
+        if message.type == "skill.converse.response":
+            if not message.data.get("result"):
+                return
+        self._response.set()
+
+    def _receive_answer(self, message):
+        utt = message.data["utterance"]
+        self._responses.append(utt)
 
     # abstract Solver methods
     def get_data(self, query, context=None):
         return {"answer": self.get_spoken_answer(query, context)}
 
-    def get_spoken_answer(self, query, context=None):
+    def get_spoken_answer(self, query, context=None, timeout=10):
         if self.hm is None:
             LOG.error("not connected to HiveMind")
             return
+        self._response.clear()
+        self._responses = []
+
         context = context or {}
         if "session" in context:
             lang = context["session"]["lang"]
@@ -41,14 +74,11 @@ def get_spoken_answer(self, query, context=None):
             lang = context.get("lang") or self.config.get("lang", "en-us")
         mycroft_msg = Message("recognizer_loop:utterance",
                               {"utterances": [query], "lang": lang})
-        msg = HiveMessage(HiveMessageType.BUS, mycroft_msg)
-        response = self.hm.wait_for_payload_response(
-            message=msg,
-            reply_type=HiveMessageType.BUS,
-            payload_type="speak",
-            timeout=20)
-        if response:
-            return response.payload.data["utterance"]
+        self.hm.emit_mycroft(mycroft_msg)
+        self._response.wait(timeout=timeout)
+        if self._responses:
+            # merge multiple speak messages into one
+            return "\n".join(self._responses)
         return None  # let next solver attempt