diff --git a/packages/discovery-provider/src/tasks/repair_audio_analyses.py b/packages/discovery-provider/src/tasks/repair_audio_analyses.py
index 2fbc696f16e..19f6e3f251a 100644
--- a/packages/discovery-provider/src/tasks/repair_audio_analyses.py
+++ b/packages/discovery-provider/src/tasks/repair_audio_analyses.py
@@ -1,15 +1,12 @@
-import json
 import random
 import re
-from datetime import datetime
-from typing import List, Optional
+from typing import List
 
 import requests
 from redis import Redis
 from sqlalchemy import or_
 from sqlalchemy.orm.session import Session
 
-from src.api_helpers import generate_signature
 from src.models.tracks.track import Track
 from src.tasks.celery_app import celery
 from src.utils import get_all_nodes
@@ -20,7 +17,7 @@
 
 REPAIR_AUDIO_ANALYSES_LOCK = "repair_audio_analyses_lock"
 DEFAULT_LOCK_TIMEOUT_SECONDS = 30 * 60  # 30 minutes
-BATCH_SIZE = 1000
+BATCH_SIZE = 5000
 
 
 def query_tracks(session: Session) -> List[Track]:
@@ -35,7 +32,7 @@ def query_tracks(session: Session) -> List[Track]:
             Track.genre != "Podcast",
             Track.genre != "Audiobooks",
         )
-        .order_by(Track.track_id.asc())
+        .order_by(Track.track_id.desc())
         .limit(BATCH_SIZE)
         .all()
     )
@@ -49,69 +46,16 @@ def select_content_nodes(redis: Redis):
     return random.sample(endpoints, min(5, len(endpoints)))
 
 
-def retrigger_audio_analysis(
-    nodes: List[str],
-    track_id: int,
-    track_cid: str,
-    upload_id: Optional[str],
-    legacy_track: bool,
-):
-    cid = ""
-    if legacy_track:
-        cid = track_cid
-        upload_id = ""
-    data = {
-        "trackId": track_id,
-        "cid": cid,
-        "timestamp": int(datetime.utcnow().timestamp() * 1000),
-        "upload_id": upload_id,
-    }
-    signature = generate_signature(data)
-    params = {
-        "signature": json.dumps(
-            {
-                "data": json.dumps(data),
-                "signature": signature,
-            }
-        )
-    }
-    for node in nodes:
-        try:
-            endpoint = (
-                f"{node}/tracks/legacy/{track_cid}/analyze"
-                if legacy_track
-                else f"{node}/uploads/{upload_id}/analyze"
-            )
-            resp = requests.post(endpoint, params=params, timeout=5)
-            if resp.status_code == 400:
-                # Should only return a 400 if content found the track_cid to not be an audio file, therefore unable to be analyzed.
-                # Quit early because all CNs will 400 for this cid.
-                logger.warning(
-                    f"repair_audio_analyses.py | attempt to trigger audio analysis for track {track_id} (track_cid: {track_cid}, audio_upload_id: {upload_id}) failed with status code {resp.status_code}: {resp.text}"
-                )
-                return
-            resp.raise_for_status()
-            return
-        except Exception:
-            # Fallback to the next node
-            continue
-    logger.warning(
-        f"repair_audio_analyses.py | failed to trigger audio analysis for track {track_id} (track_cid: {track_cid}, audio_upload_id: {upload_id}). tried {nodes}"
-    )
-
-
 def repair(session: Session, redis: Redis):
     # Query batch of tracks that are missing key or bpm and have err counts < 3 from db
     tracks = query_tracks(session)
     nodes = select_content_nodes(redis)
     num_tracks_updated = 0
-    num_analyses_retriggered = 0
     for track in tracks:
         if not track.track_cid:
             # Only analyze streamable tracks
             continue
         legacy_track = not track.audio_upload_id
-        found = False
         for node in nodes:
             try:
                 # Query random content node for the audio upload id
@@ -156,39 +100,14 @@ def repair(session: Session, redis: Redis):
             if track_updated:
                 num_tracks_updated += 1
 
-            # Failures get retried up to 3 times
-            # if (not key or not bpm) and error_count < 3:
-            #     # Trigger another audio analysis but don't bother polling for result. Will read it in next batch.
-            #     num_analyses_retriggered += 1
-            #     retrigger_audio_analysis(
-            #         nodes,
-            #         track.track_id,
-            #         track.track_cid,
-            #         track.audio_upload_id,
-            #         legacy_track,
-            #     )
             if error_count >= 3:
                 logger.warning(
                     f"repair_audio_analyses.py | Track ID {track.track_id} (track_cid: {track.track_cid}, audio_upload_id: {track.audio_upload_id}) failed audio analysis >= 3 times"
                 )
-            found = True
             break
 
-        if not found:
-            logger.warning(
-                f"repair_audio_analyses.py | failed to query audio analysis for track {track.track_id} (track_cid: {track.track_cid}, audio_upload_id: {track.audio_upload_id}). tried {nodes}. triggering analysis..."
-            )
-            num_analyses_retriggered += 1
-            retrigger_audio_analysis(
-                nodes,
-                track.track_id,
-                track.track_cid,
-                track.audio_upload_id,
-                legacy_track,
-            )
-
     logger.info(
-        f"repair_audio_analyses.py | updated {num_tracks_updated} tracks, retriggered analyses for {num_analyses_retriggered} tracks. last track ID processed: {tracks[-1].track_id}"
+        f"repair_audio_analyses.py | updated {num_tracks_updated} tracks. last track ID processed: {tracks[-1].track_id}"
     )