I am looking for a reference (or a simple proof) of the fact that a free group is sofic. The preferred dynamical definition of a sofic group seems to be that there is a sequence of finite sets $V_n$ with $|V_n|\to\infty$ and a sequence of maps $\sigma_n\colon \Gamma\to \text{Sym}(V_n)$ such that for (1) fixed $g$ and $h$, $\sigma_n(gh)(v)=\sigma_n(g)\sigma_n(h)v$ for most $v\in V_n$ as $n\to\infty$; and (2) for fixed $g\ne e$, $\sigma_n(g)v\ne v$ for most $v\in V_n$ as $n\to\infty$.
Many references that I have seen establish this property for amenable groups (where $V_n$ is taken to be a Følner sequence in $\Gamma$, and $\sigma_n(g)$ is defined to be left multiplication by $g$ where this leaves elements of $V_n$ inside $V_n$; and defined arbitrarily to make $\sigma_n(g)$ a bijection for other elements of $V_n$) and for residually finite groups (where $V_n$ is taken to be a sequence of quotients of $\Gamma$ by increasing normal subgroups of finite index).
It is quite straightforward to see that $SL(d,\mathbb Z)$, for example, is residually finite. The case of the free group $F_n$ is described in the references that I have seen (a number of lectures online, Lewis Bowen's ICM address and the book of Kerr and Li) as an "interesting technical exercise", or is deduced by embedding $F_n$ in $SL(d,\mathbb Z)$.
Does anyone know a direct proof that $F_n$ is sofic, or have a reference to such a proof?