The biggest improvement for the dollar comes from eliminating bottlenecks. The Acela does not hit any speed records. It spends most of its time running at 200 kph or below. It made its big gains through taking slow turns a little faster, eliminating stops, better acceleration and braking, and eliminating the engine change in New Haven. This cut the NYC-WAS time by about 30 minutes (with no change to top speed compared to the predecessor Metroliner), and the NYC-BOS time by about 1 hour. Analysis of schedule impact continues to show that the biggest wins will be through upgrading the very slow track and signaling on NYC to New Haven, and through replacing worn out bridges and sections that require go slow orders.
Similar improvements in California have dramatically increased capacity and ridership in both the LA-SAN traffic and OAK-SAC traffic.
Perhaps a third of the first round of grants went to this kind of project. They tend to be small, plebian projects that add sidings, straighten track, bypass bottlenecks, add overpasses, etc. Only the locals notice the improvement. These are hard sells to national politicians.
The second round seems to have done much better. The details are in their report (PDF). In addition to eliminating passenger bottlenecks, it is funding many projects that eliminate freight bottlenecks and road-rail interference. Replacing some at grade crossings with overpasses can eliminate long traffic delays on the roads and allow heavy freight traffic during commuting hours. Both sides win. The Long Beach port improvements several decades ago were the first to exploit this kind of improvement. The CREATE project around Chicago is the current largest of these coordinated projects. There are several smaller such projects also in progress and an East Coast corridor analysis looking at a huge series of bottleneck improvements for the Virgina to Massachusetts corridor of rail, road, sea, and air traffic.