Round the back, there are two paralleled speaker outputs. Anyone who has ever used a stage rig with real stereo delay and reverb will rue the fact that the power stage of the TAE is mono — an opportunity missed, perhaps, although I can see that the design choice is one that makes sense for the majority of users.
There's a headphone output, with independent level control, one use for which would surely be silent personal practice using a real amp without having to annoy anyone else. There's no 'clean' analogue line input for connecting a backing-track player so, as with the OX, you'd have to use a small mixer to create that sort of setup. The effects loop Return might look like a candidate, for a mono source at least, but it returns before the speaker-sim stage, and you wouldn't want your backing track playing back through that. What you can do, however, is input a digital signal via the USB port — the TAE can actually be instantiated as an audio interface within suitable software, such as a DAW, which means you can also record a digital signal from it. I found setting it up for recording and stereo playback in Logic Pro entirely painless, with just the initial defaults working exactly as you'd want.
The TAE offers two USB audio output streams: 'Line' and 'Pre Effect': the latter taking a mono signal from immediately after the external effects loop and thus before the internal effects and speaker sim. You can record from both simultaneously, so if you like the performance but not the sound, you can send the Pre Effect signal back to the TAE for 're-speakering', or treat it with an IR of your choice in your DAW. It's not 're-amping', of course, as the signal already includes the sound of the amp.
The USB port is also the connection to the software editor and IR loader apps (I'm not sure why the latter isn't just part of the editing software). The functional part of the editor is organised as three panels: Rigs to the left, block diagram at the top, with parameters for the selected process below. A full snapshot of (nearly) everything is a Rig, and you can make groupings of up to 15 Rigs as a Live Set, with the editor software allowing you to create up to 50 Live Sets. Rigs and Live Sets can be named, saved and backed up, along with system settings.
The analogue line outputs, on XLRs, appear as a stereo pair, plus a summed mono — the latter, a good, practical inclusion for an FOH feed, is independent of the Line Out level, but perhaps a bit hot for something that will often be plugged into a mic amp. The only 'stereo' here, however, is the delays and reverb/room effects. There's no pan facility or left/right assign to allow you to take advantage of the Direct Mix control in the software: the direct signal offers a pre-cab output that could be usefully recorded simultaneously for the same purpose as the USB clean-feed facility, but only if it could be assigned its own output. You can get an independent, pre-cab output from the effects send, however, if you are not using the loop.
Designing a Rig, you have a choice of 22 cabinet types, plus four user impulse-response slots — I am sure I am not alone in wishing there were a whole lot more of the latter. There are five microphone options, one of which is 'flat', so in terms of 'character' mics we actually have only four. The Shure SM57 and Sennheiser 421 are 'standards', particularly for stage use, but AKG's C451 and a Neumann U87 seem odd choices for alternative mics, to me. Personally, I'd much rather have something like a U67 and, ideally, a Royer R-121 ribbon. Personally, I'd never choose a 451 or a U87 as a close mic for a guitar speaker. Certainly, the latter is a popular choice for a second mic, sometimes at a bit of a distance — but that's not an option here, as you only get access to one mic at a time with the integral cabs. The user-IR slots, of course, make it possible to use double-miked IRs. Some of the integral cab collection are listed as specifically 'FOH' versions. In these, the characteristic peaks and dips of the response have been slightly smoothed out, as this was found by the designers to be preferable for live amplification, compared to the 'full-detail' versions used in a recording situation.
The Mic Position parameters are a choice of Short, Medium or Long, combined with a nine-step Distance parameter, from 2 to 10 cm, representing distance sideways from the centre of the cone. The final parameter is a virtual Room Mic with three settings: Anechoic, Small Room and Big Room. With a fixed level at each setting and no ability to record this as a separate source, I found this to be a rather limited implementation. In a live setting, you will be in a real 'small room' or possibly a 'large room', and may not want the sound of another 'virtual' one of either size imposed on your signal. In a recording context, I can't think of any reason why you'd want a fixed amount of room sound 'baked in' to your recorded signal, when you can apply precisely the right amount with a software process in your live monitoring, or in the mix, or indeed both.
The Position and Distance parameters are disabled when you activate user-IRs. I initially thought that might be because the various settings for the internal cabs were calling different IRs, but they were subsequently described to me as "filter" processes, which begs the question, why couldn't they also work with user IRs? Perhaps the format of the integral IRs has some additional data to interact with. Precisely what that format is remains invisible, as does the eventual length of the IR used, as the separate IR loader app can assimilate a variety of formats — 44.1, 48 and 96 kHz, 16- or 24-bit — and make them its own. That's a big plus if you've got a substantial collection of IRs in different formats that you'd like to try with the TAE's reactive load.
Moving up to the front of the processing chain, we have the analogue effects loop. Other than On/Off, all the settings here are in hardware: series, parallel, +4dBu/-10dBV operation, and independent ground lifts on send and return. It seems odd to me that it is mono, when most external rack effects that would use the +4 setting will have stereo return signals. I know you could patch such a processor in-line after the TAE, but then you'd have to adjust levels in two places and lose the convenient fold-down to mono for FOH.
Next up, we have a compressor stage, offering a choice of 'Rack 160D' or 'VTG Rack U', emulating a dbx 160 or a UREI 1178 (according to the documentation). Parameters are the same for both: Threshold, Ratio, Level, Attack, and Release. There's obviously no attempt to emulate the actual controls on either unit here, but both processors are perfectly functional. In fact, the '1178' is very nice, and I used a hint of it on every Rig I designed for myself. I'm not sure why it is an '1178' — the stereo version of the FET-based 1176 — as there seems to be no way for it to receive a stereo input. There's no gain-reduction meter or indicator of any kind, and in my opinion that's a regrettable omission; it wouldn't have to be a fancy VU graphic — a line of virtual LEDs would do the same job — but it really should be there in some form. The amount of compression can be difficult to judge by ear alone with a single-source signal, especially one that is already compressed to some extent, such as distorted guitar.
The delay section offers Single, Pan, Stereo, Analog and SDE-3000. Single is obviously mono, but Pan gives you a variable sub-division of the delay for the other side of the L/R pair, while Stereo is actually one channel dry and the other delayed. Analog is a bandwidth-limited mono delay, while SDE-3000 mimics Roland's flagship, mono delay unit of the mid-1980s — its 12-bit conversion and integral modulation facilities made it a firm favourite with the 'big rack' guys of that era, and you'll still find some of them in use today. The parameters vary a little between the delay types, but all have modulation. The delay-time control scaling makes it very hard to set low values using the on-screen control. A double–click gets you a numeric keypad on screen to enter a precise time, but that is not the same as a continuous adjustment made by ear. I was similarly frustrated by the scaling of the delay level and feedback controls: most users, most of the time, will be using a delay mix between five and 15 percent, with a similar or lower setting on the feedback. Yet the entirety of that range lies with the bottom 30 degrees of the available 270-degree arc of the control, while the rest of it will almost never be used other than for the odd special effect. If this were hardware, you'd choose a log pot to put more resolution at the bottom of the scale, and it is not hard to get this right in software. Just to be fair, though, I'll reiterate that I made precisely the same criticism of the OX's software app. And, in contrast, the scaling on the front-panel Speaker Out control is just perfect, with lots of fine control where you need it most.
The reverb block offers a choice of Hall 1, Hall 2, Plate, Room 1, Room 2 and Spring. Sensibly, there are separate levels available for the speaker output and the line outputs, so you don't have to send the same mix to both. High- and Low-Damping controls and up to 200ms of pre-delay form part of a comprehensive array of controls but, like the delay section, the reverb's decay time crams its most usable, 'up to three seconds' part of the scale in a quarter of the control. Not many guitar players will use a 5-10s reverb decay time, but they may well want to adjust in the range between two and three seconds by ear, using a continuous control rather than successive approximations via numeric entry.
It allowed me to greatly enjoy some of the amps I own that are now considered anti-social — and equally joyful, at the other extreme, was hearing a diminutive H&K Tubemeister 5 at stage-viable volume!
For an interface of more than just a very few controls to be used efficiently, I believe there needs to be some degree of visual or layout hierarchy. The TAE editor screens have none. The controls are all the same size, the same colour and use the same small, all-upper-case text. Some differentiation of the most frequently used controls by size, position or colour would make a difference. In hardware design, the user interface is often compromised by the needs of the circuit and limited front-panel space, but this is software. There is no reason why it can't be better at allowing the user to locate the most important and frequently used controls more efficiently.
The final stage, before we arrive back at the speaker sim, splits the signal into two paths, with an EQ in each. The two EQs can be activated separately, or ganged for joint operation. There is a choice of 10-band graphic, or a parametric with two 'true parametric' (with Q) bands, plus Low Cut and shelving Low Gain, and High Cut and shelving High Gain. There's no graphic representation of the response curve in the editor, however — something regular DAW plug-in users are now very familiar and comfortable with. Like many who grew up using studio hardware, I am happy to interpret the virtual knob positions on the TAE's parametric, but I find those who have only ever used software EQs often don't have the same instinctive sense of what the parametric numbers mean in terms of a response shape. It's not as if there's no unused screen space to do it in.
The EQ stage can also be used as a 'solo boost' facility, with up to 20dB of footswitchable, post-EQ gain available to both line and speaker outputs, depending on whether or not the EQs are ganged. This is a post-distortion boost that will result in a genuine level change, rather than just more drive, and attaching it to an EQ can make it more usable — trimming off ultra-highs and lows can help your boosted signal to still sit in a mix properly even though it is louder. However, I found I wanted to use at least some EQ to fine-tune the basic sound of every Rig that I created for myself, which meant the solo EQ option was then no longer available as a solo boost. I would urge the addition of a second EQ stage, independent of the solo option, for any future software update.