Upper tropospheric zonal winds have been observed for decades using the cloud tracking technique (Ingersol et al. 1979, 2004). Jupiter exhibits alternating prograde and retrograde jets at the boundaries of zones and belts, with maximum velocities (of ~100m/s) reached in the tropical region and decreasing velocities towards the poles. Recent Juno Micro Wave Radiometer observations have enabled deriving that these zonal winds reach depth of 3000km below the cloud top (Kaspi et al. 2018). Above the cloud top, in the stratosphere, there are no such tracers as clouds. Zonal winds can in principle be deduced from the temperature field by assuming thermal wind balance and using the tropospheric winds as a boundary condition. Using Cassini/CIRS, Flasar et al. (2004) deduced a strong prograde jet (~140 m/s) at 3mbar and at tropical latitudes in Jupiter from the thermal field. However, wind velocities cannot be inferred at equatorial latitudes from this technique, leaving this region of intense dynamics (with the Quasi-Quadrennial Oscillation ; Friedson et al. 1999) unconstrained. <P />The very high spectral resolution attained with heterodyne spectroscopy offers us the unique means to directly measure zonal winds in planetary atmospheres (e.g., Lellouch et al. 1991, Moreno et al. 2005, 2009, Cavalié et al. 2008) by observing wind-induced Doppler shifts of the spectral line cores. We have used the Atacama Large Millimeter/submillimeter Array (ALMA) in March 2017 to map the jovian stratospheric HCN emission with a high spatial resolution. From these observations, we have measured the wind velocities at the planetary limb and at millibar/submillibar levels. In this paper, we show the detection of a strong double-peaked tropical jet at +/-10° latitude that is well-correlated with its tropospheric counterpart.